Publications

2026

1. IEEE Sensors

   Multidomain Supervised Time–Frequency Feature Fusion Autoencoder for Health Indicator Extraction in Rotating Machinery

   Hanbyeol Park, Jungho Choo, Gawon Lee, and 1 more author

   IEEE Sensors Journal, 2026

   Abs DOI Bib

   Constructing health indicators (HIs) that quantitatively represent the state of a machine is essential for the condition-based maintenance of rotating machinery. Traditional physics-based HI construction methods have limited effectiveness; for example, virtual HIs integrating multiple features have limited applicability because of the manual feature extraction process that depends on expert knowledge and lack of model adaptability. Existing studies focus on single domains (raw signals) or use time–frequency domain spectrograms and fail to adequately extract meaningful frequency band patterns extended along specific axes. These limitations in representation capacity lead to a degraded performance in HI extraction. This paper proposes a novel deep learning-based HI extraction method that uses a supervised time–frequency feature fusion autoencoder for fusing multidomain features. The proposed model includes spatiotemporal feature extraction from raw signals using a multiscale convolution neural network Transformer block, extraction of temporally resolved frequency features from time–frequency maps using rectangular kernels and cross-attention mechanisms, fusion of latent representations to extract one-dimensional HI, and simultaneous reconstruction of the input signal from the extracted HI unified encoder–decoder architecture. During training, the model optimizes a composite loss function that comprises a shape constraint function and domain-specific reconstruction errors using a dynamic weight-averaging strategy. The proposed method is validated using publicly available rotating machinery datasets and demonstrates superior performances across four evaluation metrics compared with those of existing methods. This study highlights the potential of automated high-performance HI extraction for rotating machinery and offers practical contributions to industrial maintenance applications.

   @article{11570813,
     author = {Park, Hanbyeol and Choo, Jungho and Lee, Gawon and Bae, Hyerim},
     journal = {IEEE Sensors Journal},
     title = {Multidomain Supervised Time–Frequency Feature Fusion Autoencoder for Health Indicator Extraction in Rotating Machinery},
     year = {2026},
     volume = {},
     number = {},
     pages = {1-1},
     keywords = {Modeling,Timing,Frequency,Spectrogram,Degradation,Construction,Testing,Image sensors,Machinery,Convolutional neural networks,health indicator,multidomain supervised time–frequency feature fusion autoencoder,rotating machinery},
     doi = {10.1109/JSEN.2026.3702985}
   }

   Keywords: Modeling Timing Frequency Spectrogram Degradation Construction Testing Image sensors Machinery Convolutional neural networks health indicator multidomain supervised time–frequency feature fusion autoencoder rotating machinery
2. JMARST

   Synchronous Deep Reinforcement Learning for Optimized Storage Assignment of Ship Blocks

   Gawon Lee, Jaehyeon Heo, Misung Kim, and 1 more author

   Journal of Marine Science and Technology–Taiwan, 2026

   Abs DOI HTML Bib

   This paper presents a deep reinforcement learning (RL) framework for optimizing block storage allocation in shipbuilding yards. During the shipbuilding process, vessels are constructed in block units to maximize the operational efficiency. Following assembly, these blocks must be stored in limited yard spaces, creating a complex variant of the binary packing problem. This storage allocation problem is further complicated by operational constraints, including the barge capacity and transportation time restrictions. Moreover, poor storage decisions can lead to redundant block movements, which can adversely affect downstream processes and increase operational costs. To resolve this problem, a policy-gradient-based synchronous RL model was developed and an embedding layer was introduced to improve its inference performance. The proposed model was validated using both synthetic data and real-world stockyard data. The synthetic dataset was used to prove the proposed model’s efficacy and evaluate the hyper parameter value. The real-world data with operational constraints derived from actual field conditions is used for systematic comparisons with mathematical optimization. The comparison demonstrated that the proposed RL method effectively generated solutions that satisfied practical operational constraints, offering a viable alternative to traditional mathematical methods.

   @article{lee2026synchronous,
     title = {Synchronous Deep Reinforcement Learning for Optimized Storage Assignment of Ship Blocks},
     author = {Lee, Gawon and Heo, Jaehyeon and Kim, Misung and Bae, Hyerim},
     journal = {Journal of Marine Science and Technology--Taiwan},
     volume = {34},
     doi = {10.51400/2709-6998.2831},
     number = {3},
     pages = {5},
     year = {2026},
     publisher = {National Taiwan Ocean University}
   }

3. ICPR2025

   Machine Learning-Based Initial Population Generation for Quay Crane Allocation and Scheduling Problem

   Kikun Park and Hyerim Bae

   In Sustainable Systems in an Internetworked Digital World, 2026

   Abs DOI Bib

   The quay crane assignment and scheduling problem (QCASP) involves determining the number of quay cranes (QCs) to be assigned to a container vessel and establishing detailed task schedules for each crane. The QCASP is classified as an NP-hard problem, and population-based stochastic algorithms, such as genetic algorithms (GA), are commonly employed to search for optimal solutions. However, solving such a complex problem using a GA typically requires the generation and evaluation of several chromosomes, which can significantly reduce the efficiency of the search process. This study proposes a machine learning (ML)-based method for generating the initial population to address this issue. Specifically, to reduce the search space, a clustering method is introduced for tasks based on the vessel’s stowage plan, and an initial solution generation rule is formulated by transforming the problem into a traveling salesman problem. Subsequently, the initial population is evaluated, and ML techniques are employed to learn and select high-quality solutions. The proposed approach enhances search efficiency through strategic initial solution generation while also yielding practically feasible QC task assignments.

   @inproceedings{10.1007/978-3-032-19656-9_40,
     author = {Park, Kikun and Bae, Hyerim},
     editor = {Montoya-Torres, Jairo R. and Mej{\'i}a, Gonzalo E.},
     title = {Machine Learning-Based Initial Population Generation for Quay Crane Allocation and Scheduling Problem},
     booktitle = {Sustainable Systems in an Internetworked Digital World},
     year = {2026},
     publisher = {Springer Nature Switzerland},
     address = {Cham},
     doi = {https://doi.org/10.1007/978-3-032-19656-9_40},
     pages = {356--367},
     keywords = {Quay crane assignment and scheduling problem, Machine learning, Initial population generation, Genetic algorithm, Clustering, Traveling salesman problem}
   }

   Keywords: Quay crane assignment and scheduling problem Machine learning Initial population generation Genetic algorithm Clustering Traveling salesman problem
4. arXiv

   LAST-RAG: Literature-Anchored Stochastic Trajectory Retrieval-Augmented Generation for Knowledge-Conditioned Degradation Model Selection

   Hanbyeol Park and Hyerim Bae

   2026

   Abs DOI arXiv PDF Bib

   Stochastic-process-based degradation modeling is a core approach for estimating the distribution of remaining useful life (RUL); however, the selection of an appropriate stochastic process has not been sufficiently addressed. Existing model selection methods mainly rely on the statistical fit of the observed health indicator (HI) trajectory, but this approach may select a model that is inconsistent with the underlying degradation mechanism when the observation window is short or the signal is highly noisy. To address this issue, this paper proposes Literature-Anchored Stochastic Trajectory Retrieval-Augmented Generation (LAST-RAG). The proposed method uses both the observed HI trajectory and domain-specific context, and hierarchically conditions the candidate degradation model space based on theoretical and mechanical evidence retrieved from a local evidence bank. In addition, Rule-based Confidence Reasoning with Uncertain State (RCRUS) is introduced to prevent candidate models from being prematurely eliminated when hierarchical decisions are uncertain. Simulation-based experiments demonstrate that the proposed method outperforms statistical, prognostic, and uncertainty-aware baselines in both Wiener/gamma family classification and detailed degradation model classification. Ultimately, this study reframes degradation model selection from a purely statistical goodness-of-fit problem into a knowledge-conditioned decision-making problem that integrates observed data with domain knowledge.

   @misc{park2026lastragliteratureanchoredstochastictrajectory,
     title = {LAST-RAG: Literature-Anchored Stochastic Trajectory Retrieval-Augmented Generation for Knowledge-Conditioned Degradation Model Selection},
     author = {Park, Hanbyeol and Bae, Hyerim},
     year = {2026},
     archiveprefix = {arXiv},
     primaryclass = {cs.AI},
     doi = {https://doi.org/10.48550/arXiv.2605.17902},
     keywords = {Degradation model selection, Stochastic processes, Remaining useful life estimation, Retrieval-augmented generation, Uncertainty-aware decision making, Large language models}
   }

   Keywords: Degradation model selection Stochastic processes Remaining useful life estimation Retrieval-augmented generation Uncertainty-aware decision making Large language models
5. arXiv

   LoopUS: Recasting Pretrained LLMs into Looped Latent Refinement Models

   Taekhyun Park, Yongjae Lee, Dohee Kim, and 1 more author

   2026

   Abs DOI arXiv HTML Bib

   Looped computation shows promise in improving the reasoning-oriented performance of LLMs by scaling test-time compute. However, existing approaches typically require either training recurrent models from scratch or applying disruptive retrofits, which involve substantial computational costs and may compromise pretrained capabilities. To address these limitations, we introduce \textbfLooped Depth Up-Scaling (LoopUS), a post-training framework that converts a standard pretrained LLM into a looped architecture. As a key technical contribution, LoopUS recasts the pretrained LLM into an encoder, a looped reasoning block, and a decoder. It operationalizes this latent-refinement architecture through four core components: (1) block decomposition, guided by staged representation dynamics; (2) an input-dependent selective gate to mitigate hidden-state drift; (3) random deep supervision for memory-efficient learning over long recursive horizons; and (4) a confidence head for adaptive early exiting. Collectively, these mechanisms transform a standard non-looped model into a looped form while stabilizing it against both computational bottlenecks and representation collapse. Through stable latent looping, LoopUS improves reasoning-oriented performance without extending the generated traces or requiring recurrent training from scratch.

   @misc{park2026loopusrecastingpretrainedllms,
     title = {LoopUS: Recasting Pretrained LLMs into Looped Latent Refinement Models},
     author = {Park, Taekhyun and Lee, Yongjae and Kim, Dohee and Bae, Hyerim},
     year = {2026},
     keywords = {Looped computation, Large language models, Post-training framework, Latent-refinement architecture, Selective gate, Random deep supervision, Confidence head},
     doi = {https://doi.org/10.48550/arXiv.2605.11011},
     archiveprefix = {arXiv},
     primaryclass = {cs.LG},
   }

   Keywords: Looped computation Large language models Post-training framework Latent-refinement architecture Selective gate Random deep supervision Confidence head
6. arXiv

   Domain-Adaptive Health Indicator Learning with Degradation-Stage Synchronized Sampling and Cross-Domain Autoencoder

   Jungho Choo, Hanbyeol Park, Gawon Lee, and 2 more authors

   2026

   Abs arXiv PDF Bib

   The construction of high quality health indicators (HIs) is crucial for effective prognostics and health management. Although deep learning has significantly advanced HI modeling, existing approaches often struggle with distribution mismatches resulting from varying operating conditions. Although domain adaptation is typically employed to mitigate these shifts, two critical challenges remain: (1) the misalignment of degradation stages during random mini-batch sampling, resulting in misleading discrepancy losses, and (2) the structural limitations of small-kernel 1D-CNNs in capturing long-range temporal dependencies within complex vibration signals. To address these issues, we propose a domain-adaptive framework comprising degradation stage synchronized batch sampling (DSSBS) and the cross-domain aligned fusion large autoencoder (CAFLAE). DSSBS utilizes kernel change-point detection to segment degradation stages, ensuring that source and target mini-batches are synchronized by their failure phases during alignment. Complementing this, CAFLAE integrates large-kernel temporal feature extraction with cross-attention mechanisms to learn superior domain-invariant representations. The proposed framework was rigorously validated on a Korean defense system dataset and the XJTU-SY bearing dataset, achieving an average performance enhancement of 24.1% over state-of-the-art methods. These results demonstrate that DSSBS improves cross-domain alignment through stage-consistent sampling, whereas CAFLAE offers a high-performance backbone for long-term industrial condition monitoring.

   @misc{choo2026domainadaptivehealthindicatorlearning,
     title = {Domain-Adaptive Health Indicator Learning with Degradation-Stage Synchronized Sampling and Cross-Domain Autoencoder},
     author = {Choo, Jungho and Park, Hanbyeol and Lee, Gawon and Park, Yunkyung and Bae, Hyerim},
     year = {2026},
     archiveprefix = {arXiv},
     primaryclass = {cs.LG},
     keywords = {Health indicator learning, Domain adaptation, Degradation stage synchronization, Large-kernel autoencoder, Cross-attention, Industrial condition monitoring},
   }

   Keywords: Health indicator learning Domain adaptation Degradation stage synchronization Large-kernel autoencoder Cross-attention Industrial condition monitoring
7. arXiv

   Generative AI and Machine Learning Collaboration for Container Dwell Time Prediction via Data Standardization

   Minseop Kim, Takhyeong Kim, Taekhyun Park, and 2 more authors

   2026

   Abs arXiv PDF Bib

   Import container dwell time (ICDT) prediction is a key task for improving productivity in container terminals, as accurate predictions enable the reduction of container re-handling operations by yard cranes. Achieving this objective requires accurately predicting the dwell time of individual containers. However, the primary determinants of dwell time-owner information and cargo information-are recorded as unstructured text, which limits their effective use in machine learning models. This study addresses this limitation by proposing a collaborative framework that integrates generative artificial intelligence (Gen AI) with machine learning. The proposed framework employs Gen AI to standardize unstructured information into standard international codes, with dynamic re-prediction triggered by electronic data interchange state updates, enabling the machine learning model to predict ICDT accurately. Extensive experiments conducted on real container terminal data demonstrate that the proposed methodology achieves a 13.88% improvement in mean absolute error compared to conventional models that do not utilize standardized information. Furthermore, applying the improved predictions to container stacking strategies achieves up to 14.68% reduction in the number of relocations, thereby empirically validating the potential of Gen AI to enhance productivity in container terminal operations. Overall, this study provides both technical and methodological insights into the adoption of Gen AI in port logistics and its effectiveness.

   @misc{kim2026generativeaimachinelearning,
     title = {Generative AI and Machine Learning Collaboration for Container Dwell Time Prediction via Data Standardization},
     author = {Kim, Minseop and Kim, Takhyeong and Park, Taekhyun and Park, Hanbyeol and Bae, Hyerim},
     year = {2026},
     archiveprefix = {arXiv},
     primaryclass = {cs.CE},
     keywords = {Container dwell time prediction, Generative artificial intelligence, Machine learning, Data standardization, Container terminal operations,Large language models},
   }

   Keywords: Container dwell time prediction Generative artificial intelligence Machine learning Data standardization Container terminal operations Large language models
8. arXiv

   Application of Large Language Models for Container Throughput Forecasting: Incorporating Contextual Information in Port Logistics

   Minseop Kim, Jaeeun Kwon, Hanbyeol Park, and 3 more authors

   2026

   Abs arXiv PDF Bib

   Recent advancements in generative artificial intelligence (AI) have demonstrated its substantial potential in various fields. However, its application in port logistics remains underexplored. Ports are complex operational environments where diverse types of contextual information coexist, making them a promising domain for the implementation of generative AI and highlighting the urgency of related research. In this study, we applied a large language model (LLM)-a leading generative AI technique-to forecast container throughput, which is a critical challenge in port logistics. To this end, we adopted a state-of-the-art LLM approach and proposed a novel prompt structure designed to incorporate the contextual characteristics of port operations. Extensive experiments confirm the superiority of our method, showing that the proposed approach outperforms competitive benchmark models. Furthermore, additional experiments revealed that LLMs can effectively learn and utilize multiple layers of contextual information for inference in port logistics. Based on these findings, we explore the key constraints affecting LLM adoption in this domain and outline future research directions aimed at addressing them. Accordingly, we offer both technical and practical insights to support the effective deployment of generative AI in port logistics.

   @misc{kim2026applicationlargelanguagemodels,
     title = {Application of Large Language Models for Container Throughput Forecasting: Incorporating Contextual Information in Port Logistics},
     author = {Kim, Minseop and Kwon, Jaeeun and Park, Hanbyeol and Park, Kikun and Park, Taekhyun and Bae, Hyerim},
     year = {2026},
     archiveprefix = {arXiv},
     primaryclass = {cs.CE},
     keywords = {Container Throughput, Generative AI, Large Language Model, Port Logistics, Smart Port, Time Series Analysis},
   }

   Keywords: Container Throughput Generative AI Large Language Model Port Logistics Smart Port Time Series Analysis
9. Applied Sciences

   Quantitative Assessment of Consistency between IMO DCS and EU MRV Systems Using Large-Scale Operational Data

   Hyunju Lee and Hyerim Bae

   Applied Sciences, 2026

   Abs DOI PDF Bib

   This study presents a large-scale empirical comparison of operational efficiency metrics derived from the IMO Data Collection System (DCS) and the EU Monitoring, Reporting and Verification (MRV) framework. Using a matched dataset of 15,755 dual-reported vessels and over 50,000 ship-year observations from 2019 to 2024, paired non-parametric tests, effect size estimation, and agreement diagnostics were applied to assess consistency across monitoring systems. Results indicate that although statistically significant differences are detected (p< 0.001), practical differences are negligible (Cohen’sd< 0.025), with MRV-based values averaging approximately 1.4% lower Annual Efficiency Ratio (AER) and fuel intensity than DCS values. Distributional analysis confirms substantial overlap between datasets, and temporal trends show progressive convergence following the implementation of the Carbon Intensity Indicator (CII) regulation. However, pronounced vessel-type heterogeneity is observed. Flexible cargo vessels exhibit consistent efficiency improvements in EU-related voyages, whereas container ships show minimal variation and LNG carriers demonstrate indicator-dependent patterns. Overall, the findings indicate that DCS and MRV provide broadly comparable representations of operational efficiency, with observed differences primarily reflecting vessel-type-specific operational characteristics rather than structural inconsistencies in reporting systems. The study contributes a scalable statistical validation framework for cross-regulatory monitoring assessment.

   @article{lee2026quantitative,
     title = {Quantitative Assessment of Consistency between IMO DCS and EU MRV Systems Using Large-Scale Operational Data},
     author = {Lee, Hyunju and Bae, Hyerim},
     journal = {Applied Sciences},
     year = {2026},
     publisher = {MDPI},
     doi = {https://doi.org/10.3390/app16062911},
     keywords = {IMO DCS, EU MRV, Operational efficiency, Large-scale data analysis, Statistical validation, Carbon Intensity Indicator},
   }

   Keywords: IMO DCS EU MRV Operational efficiency Large-scale data analysis Statistical validation Carbon Intensity Indicator
10. ICPR25

    Process-Aware Procurement Lead Time Prediction for Shipyard Delay Mitigation

    Yongjae Lee, Eunhee Park, Daesan Park, and 3 more authors

    2026

    Abs DOI arXiv HTML PDF Bib

    Accurately predicting procurement lead time (PLT) remains a challenge in engineered-to-order industries such as shipbuilding and plant construction, where delays in a single key component can disrupt project timelines. In shipyards, pipe spools are critical components; installed deep within hull blocks soon after steel erection, any delay in their procurement can halt all downstream tasks. Recognizing their importance, existing studies predict PLT using the static physical attributes of pipe spools. However, procurement is inherently a dynamic, multi-stakeholder business process involving a continuous sequence of internal and external events at the shipyard, factors often overlooked in traditional approaches. To address this issue, this paper proposes a novel framework that combines event logs, dataset records of the procurement events, with static attributes to predict PLT. The temporal attributes of each event are extracted to reflect the continuity and temporal context of the process. Subsequently, a deep sequential neural network combined with a multi-layered perceptron is employed to integrate these static and dynamic features, enabling the model to capture both structural and contextual information in procurement. Comparative experiments are conducted using real-world pipe spool procurement data from a globally renowned South Korean shipbuilding corporation. Three tasks are evaluated, which are production, post-processing, and procurement lead time prediction. The results show a 22.6% to 50.4% improvement in prediction performance in terms of mean absolute error over the best-performing existing approaches across the three tasks. These findings indicate the value of considering procurement process information for more accurate PLT prediction.

    @misc{lee2026processawareprocurementleadtime,
      title = {Process-Aware Procurement Lead Time Prediction for Shipyard Delay Mitigation},
      author = {Lee, Yongjae and Park, Eunhee and Park, Daesan and Kim, Dongho and Choi, Jongho and Bae, Hyerim},
      year = {2026},
      journal = {International Conference on Pattern Recognition (ICPR 2025)},
      doi = {https://doi.org/10.1007/978-3-032-19656-9_1},
      archiveprefix = {arXiv},
      primaryclass = {cs.LG},
      keywords = {Procurement lead time prediction, Process mining, Time Series Forecasting, Deep learning, Shipbuilding industry}
    }

    Keywords: Procurement lead time prediction Process mining Time Series Forecasting Deep learning Shipbuilding industry
11. arXiv

    ACFormer: Mitigating Non-linearity with Auto Convolutional Encoder for Time Series Forecasting

    Gawon Lee, Hanbyeol Park, Minseop Kim, and 2 more authors

    2026

    Abs DOI arXiv PDF Bib

    Time series forecasting (TSF) faces challenges in modeling complex intra-channel temporal dependencies and inter-channel correlations. Although recent research has highlighted the efficiency of linear architectures in capturing global trends, these models often struggle with non-linear signals. To address this gap, we conducted a systematic receptive field analysis of convolutional neural network (CNN) TSF models. We introduce the "individual receptive field" to uncover granular structural dependencies, revealing that convolutional layers act as feature extractors that mirror channel-wise attention while exhibiting superior robustness to non-linear fluctuations. Based on these insights, we propose ACFormer, an architecture designed to reconcile the efficiency of linear projections with the non-linear feature-extraction power of convolutions. ACFormer captures fine-grained information through a shared compression module, preserves temporal locality via gated attention, and reconstructs variable-specific temporal patterns using an independent patch expansion layer. Extensive experiments on multiple benchmark datasets demonstrate that ACFormer consistently achieves state-of-the-art performance, effectively mitigating the inherent drawbacks of linear models in capturing high-frequency components.

    @misc{lee2026acformermitigatingnonlinearityauto,
      title = {ACFormer: Mitigating Non-linearity with Auto Convolutional Encoder for Time Series Forecasting},
      author = {Lee, Gawon and Park, Hanbyeol and Kim, Minseop and Kim, Dohee and Bae, Hyerim},
      year = {2026},
      doi = {https://doi.org/10.48550/arXiv.2601.20611},
      archiveprefix = {arXiv},
      primaryclass = {cs.LG},
      keywords = {Time Series Forecasting, Deep learning, Convolutional Neural Network, Auto Convolution, Neural Network, Local variation modeling}
    }

    Keywords: Time Series Forecasting Deep learning Convolutional Neural Network Auto Convolution Neural Network Local variation modeling
12. TRD

    How Will Arctic Shipping Emissions Evolve? A Spatiotemporal Topology-Aware Transformer Approach

    Younghwi Kim, Jeahoon Lee, Hyerim Bae, and 1 more author

    Transportation Research Part D: Transport and Environment. JCR Q1, Top 7%!!! , 2026

    Abs DOI PDF Bib

    We propose a novel spatiotemporal forecasting framework that integrates spatial clustering with a transformer-based model, termed spatiotemporal topology-aware transformer (SToT), to predict maritime emissions across the Arctic region. Using Arctic Ship Traffic Data (ASTD), we construct node-based representations through spatial clustering and jointly learn spatial correlations and temporal patterns using SToT. Experimental results demonstrate that our model outperforms existing methods by 17% in terms of root mean square error, relative standard error, and node-level Pearson correlation coefficient, validating its predictive accuracy and robustness. Based on the SToT-based forecast extending through 2030, total emissions are projected to increase by 30.8% compared with 2022 emissions, with N2O and CO2 exhibiting particularly significant growth. Furthermore, this paper discusses the environmental, scientific, and political implications of maritime emissions. The proposed framework offers a scalable and accurate solution for climate risk assessments and sustainable policy planning in the Arctic region.

    @article{Kim2025ArcticShipping,
      title = {How Will Arctic Shipping Emissions Evolve? A Spatiotemporal Topology-Aware Transformer Approach},
      journal = {Transportation Research Part D: Transport and Environment},
      volume = {151},
      pages = {105134},
      year = {2026},
      issn = {1361-9209},
      doi = {https://doi.org/10.1016/j.trd.2025.105134},
      author = {Kim, Younghwi and Lee, Jeahoon and Bae, Hyerim and Sim, Sunghyun},
      keywords = {Arctic Ship Traffic Data, Arctic Environment, Ship Emissions, Time Series Forecasting, Transformer, Trans-Arctic Shipping Route},
    }

    Keywords: Arctic Ship Traffic Data Arctic Environment Ship Emissions Time Series Forecasting Transformer Trans-Arctic Shipping Route
13. Ocean Eng.

    Deep learning framework for regional maritime collision risk assessment using CNN and Grad-CAM

    Suhyeon Jo, Gawon Lee, Sekil Park, and 4 more authors

    Ocean Engineering. JCR Q1, Top 3.08%!!! , 2026

    Abs DOI HTML Bib

    Maritime transport handles over 80% of global trade but rising traffic density and congestion have increased collision risk. Traditional assessment methods, including the ship domain and closest point-of-approach approaches, are limited to pairwise interactions and cannot capture regional traffic dynamics. To overcome these shortcomings, we propose a regional collision risk assessment framework integrating AIS-derived relative traffic representations, convolutional neural networks, and explainable artificial intelligence. A gradient-weighted class activation mapping framework generated risk influence distribution maps and a centroid radial distribution, while a novel indicator, the risk influence radius (RIR), quantified radial influence patterns. Experiments with 3,660 low-risk and 366 high-risk cases from South Korean coastal waters revealed that the Visual Geometry Group (VGG) network backbone achieved the highest performance, with an average F1-score of 0.9324, outperforming ResNeXt, EfficientNet, ResNet, and MobileNetV3 models. Statistical tests revealed a significant difference in RIR (p < 0.05) between high- and low-risk collision areas at 10 and 20 km, but not at 30 km. These results confirm RIR as an interpretable indicator of collision risk at regional scales, while highlighting its limitations at broader ranges. The framework combines accuracy with interpretability and supports practical applications such as maritime monitoring, route planning, and early warning systems.applications such as maritime monitoring, route planning, and early warning systems.

    @article{JO2026123452,
      title = {Deep learning framework for regional maritime collision risk assessment using CNN and Grad-CAM},
      journal = {Ocean Engineering},
      volume = {343},
      pages = {123452},
      year = {2026},
      issn = {0029-8018},
      doi = {https://doi.org/10.1016/j.oceaneng.2025.123452},
      author = {Jo, Suhyeon and Lee, Gawon and Park, Sekil and Heo, Jaehyeon and Kim, Dohee and Sim, Sunghyun and Bae, Hyerim},
      keywords = {Regional maritime collision risk assessment, Relative traffic representation, Risk influence distribution map, Centroid radial distribution, Deep learning, Explainable AI},
    }

    Keywords: Regional maritime collision risk assessment Relative traffic representation Risk influence distribution map Centroid radial distribution Deep learning Explainable AI
14. ICIC

    IDENTIFYING SEVERITY OF MARITIME ACCIDENTS BASED ON WEIGHTED KERNEL DENSITY ESTIMATION

    Sangmin Jo, Dohee Kim, Seongmoon Hong, and 4 more authors

    In , 2026

    Abs DOI PDF Bib

    With the growing volume of maritime logistics, coastal marine traffic environments have become increasingly congested, heightening the risk of maritime accidents.Given their severe human and environmental consequences, identifying risk hotspots and implementing proactive management strategies are essential. Maritime accident severity is not merely a post-hoc score but a practical indicator for prioritizing resources and shaping safety policies. Since real-world damage is strongly influenced by concurrent weather conditions, a valid severity index must incorporate meteorological variables. Meteorological factors such as strong winds, rough seas, strong currents, and good visibility have been widely recognized to substantially influence the severity of maritime accidents in previous studies. Nonetheless, empirical studies that systematically incorporate these factors into the construction of quantitative accident severity indices remain scarce. This study, focusing on South Korean coastal waters, proposes a framework that quantitatively derives optimal severity weights from meteorological data by minimizing prediction error (MSE) in an MLP (Multi-Layer Perceptron) model. Using this refined index, a continuous spatial risk map is generated, and SHAP (Shapley Additive Explanations) analysis identifies key influencing factors. The proposed approach strengthens the explanatory validity of the severity index and supports the transition from reactive analysis to proactive risk

    @inproceedings{jo2025identifying,
      title = {IDENTIFYING SEVERITY OF MARITIME ACCIDENTS BASED ON WEIGHTED KERNEL DENSITY ESTIMATION},
      author = {Jo, Sangmin and Kim, Dohee and Hong, Seongmoon and Park, Taekhyun and Kim, Dohee and Lee, Changdong and Bae, Hyerim},
      journal = {ICIC Express Letters},
      address = {Kitakyushu, Japan},
      organization = {ICIC International},
      doi = {10.24507/icicel.20.04.391},
      keywords = {Maritime accidents, Risk mapping, Accident severity, XAI, SHAP},
      year = {2026}
    }

    Keywords: Maritime accidents Risk mapping Accident severity XAI SHAP
15. arXiv

    Rewarding Structural Conformance of Reasoning using Process Mining

    Yongjae Lee, Taekhyun Park, Sunghyun Sim, and 1 more author

    2026

    Abs DOI arXiv Bib

    Recent advances in sparse reward policy gradient methods have enabled effective reinforcement learning (RL)-based language model post-training. However, for reasoning tasks such as mathematical problem solving, binarized outcome rewards provide limited feedback on intermediate reasoning steps. While some studies have attempted to address this issue by estimating overall reasoning quality, it remains unclear whether these rewards are reliable proxies for the quality of stepwise reasoning. In this study, we consider reasoning as a structured process and propose TACReward, the reward model that can be seamlessly integrated into sparse reward policy gradient methods without additional human annotation costs or architectural modifications. TACReward aggregates stepwise structural deviations between teacher and policy reasoning using process mining techniques, producing a scalar output reward range of [0, 1] to indicate reasoning quality. Experiments on multiple mathematical reasoning benchmarks demonstrate that integrating the TACReward into sparse reward frameworks encourages the policy model to improve the structural quality of reasoning. Consequently, this leads to consistent performance improvements over existing sparse reward frameworks.

    @misc{lee2026reasoningawareproxyrewardmodel,
      journal = {arXiv preprint},
      title = {Rewarding Structural Conformance of Reasoning using Process Mining},
      author = {Lee, Yongjae and Park, Taekhyun and Sim, Sunghyun and Bae, Hyerim},
      keywords = {Reinforcement learning, Process mining, Large Language Model, Reasoning, GRPO, Deep learning},
      year = {2026},
      archiveprefix = {arXiv},
      primaryclass = {cs.AI},
      url = {https://arxiv.org/abs/2510.25065},
      doi = {https://doi.org/10.48550/arXiv.2510.25065}
    }

    Keywords: Reinforcement learning Process mining Large Language Model Reasoning GRPO Deep learning
16. Adv.Eng.Info

    Time-series approach to vessel turnaround time forecasting using queuing-based operation indicators

    Daesan Park, Taeeon Noh, Yohan Koo, and 3 more authors

    Advanced Engineering Informatics. JCR Q1, Top 2.00%!!! , 2026

    Abs DOI PDF Bib

    As global supply chains grow increasingly complex and cargo volumes continue to rise, the accurate prediction of vessel turnaround time (VTT) has become essential for enhancing port operational efficiency. Previous studies primarily utilized static features, such as vessel specifications and container volumes, often failing to capture the dynamic nature of real-world operations. To address this limitation, we propose a forecasting framework that incorporates dynamic operation indicators (OIs) from a two-stage queuing model into time-series models. Leveraging operational data from Busan Port, our approach achieves of up to 28% improvement in mean absolute percentage error compared with existing techniques. Explainability analysis using Shapley additive explanations (SHAP) further demonstrates that OIs rank among the most influential features. By providing more reliable VTT predictions, port operators can enhance berth planning, optimize resource allocation, minimize scheduling delays, and enhance overall port efficiency. These results highlight the practical value of the framework for data-driven decision-making in port operations.

    @article{PARK2026103974,
      title = {Time-series approach to vessel turnaround time forecasting using queuing-based operation indicators},
      journal = {Advanced Engineering Informatics},
      volume = {69},
      pages = {103974},
      year = {2026},
      issn = {1474-0346},
      doi = {https://doi.org/10.1016/j.aei.2025.103974},
      author = {Park, Daesan and Noh, Taeeon and Koo, Yohan and Lee, Hyeonjik and Kim, Dowon and Bae, Hyerim},
      keywords = {Maritime transport, Vessel turnaround time, Port operations, Queueing theory, Time Series Forecasting},
    }

    Keywords: Maritime transport Vessel turnaround time Port operations Queueing theory Time Series Forecasting

2025

1. SSRN

   A BERT-based Denoising Framework for Inter-Company Similarity Analysis Using 10-K Reports

   Sangmin Jo, Taehyeon Kim, Jeongho Park, and 3 more authors

   2025

   Abs HTML Bib

   The identification of similar profiled companies is essential in financial practices, including portfolio design, asset pricing, and risk attribution. Recently, studies have been increasingly turning to text-based methods that analyze the actual business descriptions of companies for similarity analysis. These studies aim to uncover industry structures through the semantic content embedded in the businesses themselves for similarity analysis. However, prior studies directly embedded the given texts for analysis, which might include content unrelated to similarity analysis, such as financial disclosures, regulatory notices, or public relations messages. These unrelated contents could degrade the performance of the analysis. To overcome this limitation, this paper proposes a method that fine-tunes the Bidirectional Encoder Representations from Transformers (BERT) model to extract sentences with low business relevance. Based on this extraction, the method reconstructs the text accordingly. Specifically, 10-K annual reports are utilized to demonstrate the effectiveness of the proposed method in peer group selection and industry clustering tasks. The performances are evaluated based on stock correlation coefficients compared to benchmark methods, along with statistical significance testing. In addition, the proposed method shows effectiveness across various text embedding techniques, indicating its generalizability in the broader context of text analysis. Finally, the proposed method interprets the similarity results using retrieval-augmented generation. This helps provide users with the underlying causes and structures that give rise to intercompany …

   @misc{Jo2025BERTDenoising,
     title = {A BERT-based Denoising Framework for Inter-Company Similarity Analysis Using 10-K Reports},
     author = {Jo, Sangmin and Kim, Taehyeon and Park, Jeongho and Yu, Jihyeon and Bae, Hyerim and Chung, Jihoon},
     year = {2025},
     journal = {SSRN},
     keywords = {Industry Clustering, Peer Group Selection, Text-based Analysis, 10-K annual reports, BERT, Large Language Models}
   }

   Keywords: Industry Clustering Peer Group Selection Text-based Analysis 10-K annual reports BERT Large Language Models
2. MARTRA

   Knowledge transfer Q-learning for vessel route planning using automatic identification system-derived expert trajectories

   Hyunju Lee, Kikun Park, and Hyerim Bae

   Maritime Transport Research, 2025

   Abs DOI HTML Bib

   Traditional route recommendation systems optimize navigation paths using environmental variables such as weather and sea conditions, but often fail to account for real-world factors encountered by mariners. To address this gap, this study proposes a knowledge transfer Q-learning (KT-QL) algorithm, a reinforcement learning method built upon the Q-learning framework. The proposed KT-QL algorithm integrates expert trajectory knowledge derived from Automatic Identification System data into the Q-learning process, enabling the agent to combine trial-and-error exploration with data-driven guidance. Experimental results show that KT-QL reduces Hausdorff distances by approximately 39 % compared with conventional reinforcement learning and traditional search methods, and enhances fuel consumption prediction accuracy by approximately 2 %. These findings highlight the potential of KT-QL to enhance maritime operational efficiency, safety, and environmental sustainability.

   @article{LEE2025100142,
     title = {Knowledge transfer Q-learning for vessel route planning using automatic identification system-derived expert trajectories},
     journal = {Maritime Transport Research},
     volume = {9},
     pages = {100142},
     year = {2025},
     issn = {2666-822X},
     doi = {https://doi.org/10.1016/j.martra.2025.100142},
     author = {Lee, Hyunju and Park, Kikun and Bae, Hyerim},
     keywords = {Reinforcement learning, Vessel trajectory prediction, Transfer learning, AIS data, Maritime navigation optimization}
   }

   Keywords: Reinforcement learning Vessel trajectory prediction Transfer learning AIS data Maritime navigation optimization
3. IEEE Access

   Continual Learning Approach with Adaptive Memory Mechanism for Predictive Process Monitoring in Dynamic Business Environments

   Alif Nur Iman, Imam Mustafa Kamal, Dohee Kim, and 1 more author

   IEEE Access, 2025

   Abs DOI Bib

   Predictive process monitoring (PPM) faces challenges when process patterns evolve over time, as prediction models experience performance degradation due to concept drift while losing accuracy on previously learned process variants through catastrophic forgetting. This study addresses these challenges in outcome-oriented PPM by integrating continual learning principles with process mining domain knowledge. We propose an adaptive memory mechanism that extracts process structure information through activity relationship analysis, categorizing relationships between activities as "Always Follows," "Sometimes Follows," or "Never Follows." The mechanism computes relation entropy vectors to characterize process patterns and uses cosine similarity between current and historical process patterns to adaptively weight memory buffers during model training. This allows the model to weight historically relevant process knowledge based on pattern similarity. We evaluate the mechanism using stability, plasticity, and trade-off metrics on the Business Process Intelligence Challenge 2015 dataset across multiple temporal splits. Results show that the adaptive memory mechanism achieves stability of 0.87 ± 0.06, plasticity of 0.84 ± 0.11, and trade-off of 0.85 ± 0.08, achieving competitive performance compared to established baseline methods. The mechanism provides explainability through visualizations that demonstrate how the model weights different memory buffers during training. Validation using concept drift detection techniques confirms that the mechanism’s behavior aligns with observable process changes, indicating adaptation to evolving process patterns while reducing catastrophic forgetting.

   @article{11264535,
     author = {Iman, Alif Nur and Kamal, Imam Mustafa and Kim, Dohee and Bae, Hyerim},
     journal = {IEEE Access},
     title = {Continual Learning Approach with Adaptive Memory Mechanism for Predictive Process Monitoring in Dynamic Business Environments},
     year = {2025},
     volume = {},
     number = {},
     pages = {1-1},
     keywords = {Predictive process monitoring, Outcome-Oriented Prediction, Process mining, Continual learning, Concept drift, Deep learning},
     doi = {10.1109/ACCESS.2025.3636194}
   }

   Keywords: Predictive process monitoring Outcome-Oriented Prediction Process mining Continual learning Concept drift Deep learning
4. INFORMS2025

   Generative AI-Driven Stochastic Maritime Route Optimization for Fuel Efficiency under Environment Uncertainty

   Hyunha Lee, Dohee Kim, Yongjae Lee, and 1 more author

   In 2025 INFORMS Annual Meeting, 2025

   Bib

   @inproceedings{lee2025GenerativeAI,
     title = {Generative AI-Driven Stochastic Maritime Route Optimization for Fuel Efficiency under Environment Uncertainty},
     author = {Lee, Hyunha and Kim, Dohee and Lee, Yongjae and Bae, Hyerim},
     keywords = {Deep learning, Generative AI, Optimization, AIS data},
     booktitle = {2025 INFORMS Annual Meeting},
     year = {2025}
   }

   Keywords: Deep learning Generative AI Optimization AIS data
5. INFORMS2025

   Synthetic–Empirical Log Fusion for Deep Learning Prediction of Outpatient Queue Dynamics

   Daesan Park, Dohee Kim, and Hyerim Bae

   In 2025 INFORMS Annual Meeting, 2025

   Abs Bib

   Classical queueing formulas provide closed-form steady-state metrics and clear managerial insights into system sensitivity under stable loads; however, their analytical reach diminishes when demand spikes or service rules change. Discrete-event simulation (DES) accurately reproduces such dynamics, but repeated runs and calibration hinder real-time use. Recent deep learning predictors enhance speed, but most are trained only on synthetic DES episodes—introducing model bias—or solely on sparse field logs, which leaves them insufficiently exposed to rare transients and measurement noise. To address these limitations, we propose a data fusion framework that provides an off-the-shelf deep learning predictor with both synthetic and empirical samples. A configurable simulator generates both routine and stress scenarios. Each episode delivers queue length paths and moments of waiting time, while the clinic logs contribute the same indicators under routine conditions. Within this framework, we systematically adjust the ratio of synthetic to empirical data to train the predictor and analyze how this ratio balances steady-state accuracy against responsiveness under stress. The proposed framework is validated using outpatient clinic data to demonstrate its practical applicability while still offering rapid inference for operational decision support. Results demonstrate (i) how prediction error varies with log share across routine and stress conditions, (ii) that fusion training remains resilient to measurement noise and log sparsity, and (iii) offers practical guidance for selecting simulation-to-log ratios—uniting the strengths of analytic formulas, simulation, and deep learning.

   @inproceedings{park2025Synthetic,
     title = {Synthetic–Empirical Log Fusion for Deep Learning Prediction of Outpatient Queue Dynamics},
     author = {Park, Daesan and Kim, Dohee and Bae, Hyerim},
     keywords = {Deep learning, Queueing theory, Discrete Event Simulation, Data fusion, Healthcare operations, Queue dynamics prediction},
     booktitle = {2025 INFORMS Annual Meeting},
     year = {2025}
   }

   Keywords: Deep learning Queueing theory Discrete Event Simulation Data fusion Healthcare operations Queue dynamics prediction
6. SCI2025

   Trajectory Feature Enhanced Spatio Temporal Anomaly Detection for Automated Guided Vehicles

   Jungho Choo, Minseop Kim, Hanbyeol Park, and 1 more author

   In The First Symposium on Computational Intelligence, 2025

   Bib

   @inproceedings{choo2025Trajectory,
     title = {Trajectory Feature Enhanced Spatio Temporal Anomaly Detection for Automated Guided Vehicles},
     author = {Choo, Jungho and Kim, Minseop and Park, Hanbyeol and Bae, Hyerim},
     keywords = {Automated Guided Vehicles, Anomaly detection, Time Series Forecasting, Trajectory features, Deep learning},
     booktitle = {The First Symposium on Computational Intelligence},
     year = {2025}
   }

   Keywords: Automated Guided Vehicles Anomaly detection Time Series Forecasting Trajectory features Deep learning
7. SCI2025

   An Integrated Framework for Probabilistic Standard Route Extraction and Modern-TCN-Based Unsupervised Anomaly Detection in AIS Data

   Jaehyeon Heo, Minseop Kim, Hanbyeol Park, and 1 more author

   In The First Symposium on Computational Intelligence, 2025

   Bib

   @inproceedings{choo2025Trajectorz,
     title = {An Integrated Framework for Probabilistic Standard Route Extraction and Modern-TCN-Based Unsupervised Anomaly Detection in AIS Data},
     author = {Heo, Jaehyeon and Kim, Minseop and Park, Hanbyeol and Bae, Hyerim},
     keywords = {Anomaly detection, AIS data, Standard route extraction, Deep learning},
     booktitle = {The First Symposium on Computational Intelligence},
     year = {2025}
   }

   Keywords: Anomaly detection AIS data Standard route extraction Deep learning
8. LOGMS

   A Real-Time Dark Vessel Detection System Using Multimodal Fusion

   Kyunghoon Lee, Seonghan Kim, Hyerim Bae, and 1 more author

   In 13th International Conference on Logistics and Maritime Systems, 2025

   Abs Bib

   Illegal maritime operations, including AIS-off navigation and dark vessel activities, present persistent challenges to maritime domain awareness. This study proposes a real-time multimodal fusion framework that integrates X-band radar tracks and Automatic Identification System (AIS) data for robust detection and characterization of dark vessels. Temporal asynchrony between radar (3–4 s) and AIS (10–180 s) sampling is addressed through a three-stage alignment strategy combining nearest assignment, linear interpolation, and geodesic-based motion prediction. Candidate associations are refined via temporal, spatial, and angular gating, with final matches optimized by a composite cost function. A systematic hyperparameter search identifies an optimal configuration ensuring both matching reliability and spatial accuracy. Subsequently, unmatched radar tracks are processed through a six-stage rule-based checker—covering trajectory continuity, speed consistency, and update regularity—to filter out physically implausible motions. Applied to real maritime data collected near Busan Port (August 1, 2025), the system successfully identifies dark-vessel candidates with a 0.744 matching rate and  106 m 90th-percentile error across 32,770 radar detections and 198 vessels. Finally, trajectories are projected into radar coordinates and rasterized for sensor-level validation, bridging the gap between traditional rule-based tracking and learning-based multimodal vessel classification. The proposed approach enables reliable dark-vessel monitoring even under limited AIS coverage, providing a scalable foundation for intelligent maritime surveillance systems.

   @inproceedings{lee2025darkvessel,
     keywords = {Dark vessel detection, Multimodal fusion, AIS data, Radar tracking, Maritime surveillance, Deep learning},
     title = {A Real-Time Dark Vessel Detection System Using Multimodal Fusion},
     author = {Lee, Kyunghoon and Kim, Seonghan and Bae, Hyerim and Lee, Changdong},
     booktitle = {13th International Conference on Logistics and Maritime Systems},
     year = {2025}
   }

   Keywords: Dark vessel detection Multimodal fusion AIS data Radar tracking Maritime surveillance Deep learning
9. LOGMS

   ALMA: Adaptive Lag Moment Alignment for Domain-Adaptive Time Series Forecasting of Ship Fuel Consumption

   Jungho Choo, Dohee Kim, Hyunha Lee, and 2 more authors

   In , 2025

   Abs Bib

   To support sustainable maritime operations, international organizations have enforced stricter environmental regulations, highlighting the need for accurate ship fuel consumption forecasting. However, existing models face generalization challenges due to limited historical data, especially for newly deployed vessels. We propose Adaptive Lag Moment Alignment (ALMA), a domain adaptation framework that transfers knowledge from data-rich source ships to target ships with only a single recorded voyage. ALMA introduces Lag Moment Discrepancy (LMD) to align temporal characteristics between domains and combines it with prediction loss. To stabilize training and balance objectives, we incorporate Dynamic Weight Averaging (DWA) into the loss function. Experimental results demonstrate that ALMA significantly reduces domain gaps and improves prediction accuracy under data-scarce conditions, offering a robust solution for maritime fuel forecasting in support of global sustainability efforts.

   @inproceedings{choo2025ALMA,
     keywords = {Transfer learning, Ship fuel consumption, Time Series Forecasting, Deep learning, Maritime sustainability, Domain adaptation},
     title = {ALMA: Adaptive Lag Moment Alignment for Domain-Adaptive Time Series Forecasting of Ship Fuel Consumption},
     author = {Choo, Jungho and Kim, Dohee and Lee, Hyunha and Kim, Minhee and Bae, Hyerim},
     journal = {13th International Conference on Logistics and Maritime Systems},
     year = {2025}
   }

   Keywords: Transfer learning Ship fuel consumption Time Series Forecasting Deep learning Maritime sustainability Domain adaptation
10. ICIC

    LURKER: Language models for Unveiling maritime Risks via Knowledge-driven Event forecasting and Reasoning

    Seongmoon Hong, Taekhyun Park, Dohee Kim, and 2 more authors

    In 19th International Conference on Innovative Computing, Information and Control (ICICIC 2025), 2025

    Abs DOI Bib

    Maritime accident risk prediction has highlighted the limitations of purely statistical classifiers and numerical scoring schemes, driving demand for scenario-driven analytical methods that can elucidate both the context and progression of incidents. In response, we propose LURKER (Language models for Unveiling maritime Risks via Knowledge-driven Event forecasting and Reasoning), an explainable framework that leverages the reasoning capabilities of LLMs (LLMs) to analyze maritime accident severity. LURKER quantifies accident risk by integrating a tree-based classification model with isotonic regression for probability calibration, then generates dynamic scenarios through a prompt-engineering module that summarizes tribunal reports and incorporates real-time data. By employing few-shot learning and chain-of-thought prompting, LURKER produces context-rich descriptions of causal factors, incident progression, and potential impacts. This approach transcends purely numerical analysis, offering comprehensive explanations of accident causes and consequences, and functions as an intelligent decision-support tool. Case-based experiments integrating tribunal reports, AIS vessel-track data, and meteorological observations demonstrate LURKER’s feasibility and practical utility as a severity-prediction tool for operational maritime safety management.

    @inproceedings{hong2025LURKER,
      title = {LURKER: Language models for Unveiling maritime Risks via Knowledge-driven Event forecasting and Reasoning},
      author = {Hong, Seongmoon and Park, Taekhyun and Kim, Dohee and Jo, Sangmin and Bae, Hyerim},
      booktitle = {19th International Conference on Innovative Computing, Information and Control (ICICIC 2025)},
      address = {Kitakyushu, Japan},
      organization = {ICIC International},
      doi = {10.24507/icicel.20.05.543},
      keywords = {Maritime accidents, Risk mapping, Accident severity, XAI, SHAP},
      year = {2025}
    }

    Keywords: Maritime accidents Risk mapping Accident severity XAI SHAP
11. Comput.Ind.Eng.

    Optimizing yard truck deployment at container terminals: a machine learning-enhanced non-dominated sorting genetic algorithm

    Kikun Park, Kihun Kim, Minseop Kim, and 1 more author

    Computers & Industrial Engineering. JCR Q1, Top 14.29%!!! , 2025

    Abs DOI PDF Bib

    The continuous rise in global container traffic has necessitated more efficient operational strategies at container terminals (CTs). However, developing effective strategies requires mitigating delays in handling equipment operations. Yard trucks (YTs) are pivotal for transporting containers between quay cranes (QCs) and yard cranes (YCs). A YT shortage can cause operational delays for both QCs and YCs, extending the makespan of main-vessel operations at the berth. Conversely, deploying an excessive number of YTs eliminates these delays but incurs higher operational costs. Consequently, determining the optimal number of YTs constitutes a bi-objective optimization problem involving a tradeoff between minimizing the operational makespan and associated YT deployment costs. This study introduces a machine learning (ML)-enhanced non-dominated sorting genetic algorithm 2 (NSGA-2) approach designed to address this challenge by minimizing both the makespan and deployment costs. The proposed method uses ML models to predict fitness scores and efficiently guide the search toward optimal solutions, while also assisting in generating, selecting, and mutating promising solutions during the optimization process. The effectiveness of ML-enhanced NSGA-2 was validated using empirical data from Busan Port Terminal in South Korea. The results highlight the superior performance and practical applicability of the proposed approach, which achieved a significant reduction of 16% in deployed YTs compared to existing strategies.

    @article{PARK2025111610,
      title = {Optimizing yard truck deployment at container terminals: a machine learning-enhanced non-dominated sorting genetic algorithm},
      journal = {Computers & Industrial Engineering},
      pages = {111610},
      year = {2025},
      issn = {0360-8352},
      doi = {https://doi.org/10.1016/j.cie.2025.111610},
      author = {Park, Kikun and Kim, Kihun and Kim, Minseop and Bae, Hyerim},
      keywords = {Container terminal operations, Yard truck deployment, Genetic algorithm, Multi-objective optimization, Machine learning},
    }

    Keywords: Container terminal operations Yard truck deployment Genetic algorithm Multi-objective optimization Machine learning
12. arXiv

    IConv: Focusing on Local Variation with Channel Independent Convolution for Multivariate Time Series Forecasting

    Gawon Lee, Hanbyeol Park, Minseop Kim, and 2 more authors

    2025

    Abs arXiv Bib

    Real-world time-series data often exhibit non-stationarity, including changing trends, irregular seasonality, and residuals. In terms of changing trends, recently proposed multi-layer perceptron (MLP)-based models have shown excellent performance owing to their computational efficiency and ability to capture long-term dependency. However, the linear nature of MLP architectures poses limitations when applied to channels with diverse distributions, resulting in local variations such as seasonal patterns and residual components being ignored. However, convolutional neural networks (CNNs) can effectively incorporate these variations. To resolve the limitations of MLP, we propose combining them with CNNs. The overall trend is modeled using an MLP to consider long-term dependencies. The CNN uses diverse kernels to model fine-grained local patterns in conjunction with MLP trend predictions. To focus on modeling local variation, we propose IConv, a novel convolutional architecture that processes the temporal dependency channel independently and considers the inter-channel relationship through distinct layers. Independent channel processing enables the modeling of diverse local temporal dependencies and the adoption of a large kernel size. Distinct inter-channel considerations reduce computational cost. The proposed model is evaluated through extensive experiments on time-series datasets. The results reveal the superiority of the proposed method for multivariate time-series forecasting.

    @misc{lee2025iconv,
      keywords = {Time Series Forecasting, Deep learning, Channel independent convolution, Multivariate forecasting, Neural Network, Local variation modeling},
      title = {IConv: Focusing on Local Variation with Channel Independent Convolution for Multivariate Time Series Forecasting},
      author = {Lee, Gawon and Park, Hanbyeol and Kim, Minseop and Kim, Dohee and Bae, Hyerim},
      journal = {arXiv preprint arXiv:2509.20783},
      year = {2025},
    }

    Keywords: Time Series Forecasting Deep learning Channel independent convolution Multivariate forecasting Neural Network Local variation modeling
13. Authorea

    Forecasting Hourly Container Throughput with EDI-Derived Factors and a Decomposition Pairwise Mixer

    Minseop Kim, Kikun Park, Hanbyeol Park, and 3 more authors

    Authorea Preprints, 2025

    Abs PDF Bib

    The continuous growth in global trade and rising uncertainty have driven extensive research on container throughput (CT) forecasting at monthly and annual scales. Recently, the demand for high-resolution CT forecasting has emerged to support real-time decision-making by practitioners. However, this presents challenges in effectively utilizing realtime port operation information and managing high-noise time series data. To address these limitations, this study introduces a preprocessing strategy and novel neural network architecture for forecasting hourly CT with a 72-hour horizon, an approach not attempted in previous studies. In the preprocessing strategy, we examined the electronic data interchange (EDI) system and extracted key factors affecting CT. Additionally, we present a novel neural network architecture—the decomposition pairwise mixer (DPM)—which is designed to effectively capture interactions among these factors. Extensive experiments were conducted with real operational data from the port of Busan in Korea, the world’s sixth-busiest port. The results reveal two key findings: derived key factors enhance the forecasting performance, and the proposed method outperforms state-ofthe-art time-series forecasting models. These findings suggest that the proposed method is not limited to long-term planning but can also substantially enhance short-term operational decision-making, thereby improving port operation efficiency.

    @article{kim2025forecasting,
      keywords = {Container throughput forecasting, EDI data, Decomposition pairwise mixer, Container terminal operations, Time Series Forecasting, Deep learning},
      title = {Forecasting Hourly Container Throughput with EDI-Derived Factors and a Decomposition Pairwise Mixer},
      author = {Kim, Minseop and Park, Kikun and Park, Hanbyeol and Lee, Kyunghoon and Kim, Kihun and Bae, Hyerim},
      journal = {Authorea Preprints},
      year = {2025},
      publisher = {Authorea}
    }

    Keywords: Container throughput forecasting EDI data Decomposition pairwise mixer Container terminal operations Time Series Forecasting Deep learning
14. IEEE Access

    Trend-fluctuation correlated attention unit for remaining useful life prediction

    Hanbyeol Park, Sunghyun Sim, Yunkyung Park, and 1 more author

    IEEE Access, 2025

    Abs DOI HTML Bib

    Health indicators (HIs) are measures that consolidate complex machinery characteristics into a single dimension and have been extensively studied for predicting the remaining useful life (RUL). Recent research has proposed HI-prediction models based on recursive cell structures that demonstrate promising performance. However, recent cell structures have limitations, such as saturation in hidden states and difficulty learning interdependent patterns among various components. Accurate RUL predictions require minimizing error accumulation in the recursive HI-prediction process, which can be achieved using a cell structure capable of capturing interdependent patterns. This study proposes a new cell structure termed the trend-fluctuation correlated attention unit (TFCAU) that employs a gate mechanism to capture interactions between divided trends and fluctuations in HI and an attention mechanism for robust long-term predictions. The superiority of the TFCAU was validated using a bearing dataset from the PRONOSTIA platform and a real-world dataset from the Korean weapons system. Comparative experiments using six sub-datasets demonstrated that TFCAU outperformed the state-of-the-art model—the gated transient fluctuation dual-attention unit—achieving a 169.37% improvement in the score metric and a reduction of 49.93 units in the mean-absolute error. An extensive ablation study experimentally confirmed that the TFCAU effectively captured the interdependent characteristics between trends and fluctuations. TFCAU delivered excellent performance for both rotating and non-rotating systems, suggesting its potential applicability across multiple operational scenarios.

    @article{11153658,
      author = {Park, Hanbyeol and Sim, Sunghyun and Park, Yunkyung and Bae, Hyerim},
      journal = {IEEE Access},
      title = {Trend-fluctuation correlated attention unit for remaining useful life prediction},
      year = {2025},
      volume = {13},
      number = {},
      pages = {164486-164507},
      keywords = {Health Indicator, Remaining Useful Life Prediction, Time series decomposition, Deep learning, Predictive maintenance, Temporal Dependencies},
      doi = {10.1109/ACCESS.2025.3607444}
    }

    Keywords: Health Indicator Remaining Useful Life Prediction Time series decomposition Deep learning Predictive maintenance Temporal Dependencies
15. BPM

    Multi-task Trained Graph Neural Network for Business Process Anomaly Detection with a Limited Number of Labeled Anomalies

    Yongjae Lee, Dohee Kim, Donghwan Kim, and 1 more author

    In International Conference on Business Process Management, 2025

    Abs DOI HTML Bib

    Anomaly detection in business processes is crucial to prevent erroneous organizational decision-making. Existing anomaly detection methods often assume access to a clean event log without anomalies, a scenario that is impractical in the real world. To solve this problem, MuGAD is proposed, a novel framework that detects anomalies in scenarios where only a limited number of labeled anomalies are available. By converting traces into a newly designed graph structure, the framework captures the control flow, order of events, and attribute information, facilitating the effective use of graph neural networks. Furthermore, MuGAD uses a two-stage training procedure designed to (1) reduce the risks posed by unlabeled traces and (2) enable the detection of anomalous traces and events. Comparative experiments are conducted on five real-life event logs to validate our proposed framework. The experimental results indicate that MuGAD outperforms the existing methods and provides better detection accuracy along with better interpretability in terms of F1-score.

    @inproceedings{lee2025multi,
      doi = {https://doi.org/10.1007/978-3-032-02867-9_22},
      keywords = {Anomaly detection, GNN, Multi-task learning, Process mining, Semi-supervised learning},
      title = {Multi-task Trained Graph Neural Network for Business Process Anomaly Detection with a Limited Number of Labeled Anomalies},
      author = {Lee, Yongjae and Kim, Dohee and Kim, Donghwan and Bae, Hyerim},
      booktitle = {International Conference on Business Process Management},
      pages = {361--378},
      year = {2025}
    }

    Keywords: Anomaly detection GNN Multi-task learning Process mining Semi-supervised learning
16. IEEE BigData

    JustDense: Just using Dense instead of Sequence Mixer for Time Series analysis

    Taekhyun Park, Yongjae Lee, Daesan Park, and 2 more authors

    In 2025 IEEE International Conference on Big Data (BigData), 2025

    Abs DOI arXiv Bib

    Sequence and channel mixers, the core mechanism in sequence models, have become the de facto standard in time series analysis (TSA). However, recent studies have questioned the necessity of complex sequence mixers, such as attention mechanisms, demonstrating that simpler architectures can achieve comparable or even superior performance. This suggests that the benefits attributed to complex sequencemixers might instead emerge from other architectural or optimization factors. Based on this observation, we pose a central question: Are common sequence mixers necessary for time-series analysis? Therefore, we propose JustDense, an empirical study that systematically replaces sequence mixers in various well-established TSA models with dense layers. Grounded in the MatrixMixer framework, JustDense treats any sequence mixer as a mixing matrix and replaces it with a dense layer. This substitution isolates the mixing operation, enabling a clear theoretical foundation for understanding its role. Therefore, we conducted extensive experiments on 29 benchmarks covering five representative TSA tasks using seven state-of-the-art TSA models to address our research question. The results show that replacing sequence mixers with dense layers yields comparable or even superior performance. In the cases where dedicated sequence mixers still offer benefits, JustDense challenges the assumption that "deeper and more complex architectures are inherently better" in TSA.

    @inproceedings{park2025justdense,
      keywords = {Time Series Forecasting, Sequence mixer, Dense layer, MatrixMixer, Neural Network, Deep learning},
      title = {JustDense: Just using Dense instead of Sequence Mixer for Time Series analysis},
      booktitle = {2025 IEEE International Conference on Big Data (BigData)},
      author = {Park, Taekhyun and Lee, Yongjae and Park, Daesan and Kim, Dohee and Bae, Hyerim},
      doi = {10.1109/BigData66926.2025.11401854},
      pages = {1172-1184},
      year = {2025}
    }

    Keywords: Time Series Forecasting Sequence mixer Dense layer MatrixMixer Neural Network Deep learning
17. J. Forecasting

    Long-term forecasting of maritime economics index using time-series decomposition and two-stage attention

    Dohee Kim, Eunju Lee, Imam Mustafa Kamal, and 1 more author

    Journal of Forecasting. JCR Q1, Top 16.17%!!! , 2025

    Abs HTML PDF Bib

    

    Forecasting the maritime economics index, including container volume and Baltic Panamax Index, is essential for long-term planning and decision-making in the shipping industry. However, studies on container volume prediction are not sufficient, and the bulk freight index has highly fluctuating characteristics, which pose a challenge in long-term prediction. This study proposes a new hybrid framework for the long-term prediction of the maritime economics index. The framework consists of time-series decomposition to break down a time-series into several components (trend, seasonality, and residual), a two-stage attention mechanism that prioritizes important variables to increase long-term prediction accuracy and a long short-term memory network that predicts and combines all components to derive the final predictive outcome. Extensive experiments are conducted using the container volume data, bulk freight index data, and various external variables. The proposed framework achieved a better predictive performance than existing time-series methods, including conventional machine learning and deep learning-based models, in the long-term prediction of container volume and the Baltic Panamax Index. Hence, the proposed method can help in decision-making through accurate long-term predictions of the maritime economics index.

    @article{kim2025long,
      keywords = {Maritime economics forecasting, Time series decomposition, Attention mechanism, LSTM, Container volume prediction, Time Series Forecasting},
      title = {Long-term forecasting of maritime economics index using time-series decomposition and two-stage attention},
      author = {Kim, Dohee and Lee, Eunju and Kamal, Imam Mustafa and Bae, Hyerim},
      journal = {Journal of Forecasting},
      volume = {44},
      number = {1},
      pages = {153--172},
      year = {2025},
      publisher = {Wiley Online Library},
    }

    Keywords: Maritime economics forecasting Time series decomposition Attention mechanism LSTM Container volume prediction Time Series Forecasting
18. Appl. Intell.

    Angular triangle distance for ordinal metric learning

    Imam Mustafa Kamal, Hyerim Bae, and Ling Liu

    Applied Intelligence, 2025

    PDF Bib

    @article{kamal2025angular,
      title = {Angular triangle distance for ordinal metric learning},
      author = {Kamal, Imam Mustafa and Bae, Hyerim and Liu, Ling},
      keywords = {Ordinal metric learning, Angular distance, Deep learning, Representation learning, Distance metric},
      journal = {Applied Intelligence},
      volume = {55},
      number = {11},
      pages = {789},
      year = {2025},
      publisher = {Springer}
    }

    Keywords: Ordinal metric learning Angular distance Deep learning Representation learning Distance metric
19. Manufacturing Systems

    Comprehensive issue identification for manufacturing data analytics implementation: Systematic literature review and case studies

    Sa-Eun Park, Sangjae Lee, Hyerim Bae, and 5 more authors

    Journal of Manufacturing Systems. JCR Q1, Top 0.94%!!! , 2025

    DOI HTML Bib

    @article{park2025comprehensive,
      title = {Comprehensive issue identification for manufacturing data analytics implementation: Systematic literature review and case studies},
      author = {Park, Sa-Eun and Lee, Sangjae and Bae, Hyerim and Kim, Kihun and Kang, Eung-Jun and Kim, Jae-Sung and Park, Yong-Min and Park, Min-Ji},
      keywords = {Manufacturing data analytics, Systematic literature review, Industry 4.0, Smart manufacturing, Data-driven decision making},
      journal = {Journal of Manufacturing Systems},
      volume = {82},
      pages = {42--59},
      issn = {0278-6125},
      doi = {https://doi.org/10.1016/j.jmsy.2025.05.006},
      year = {2025},
      publisher = {Elsevier}
    }

    Keywords: Manufacturing data analytics Systematic literature review Industry 4.0 Smart manufacturing Data-driven decision making
20. Eng. Optim.

    Contrastive loss with dynamic margin for rubber-recipe physical-property prediction

    Hanbyeol Park, Kihun Kim, Kikun Park, and 3 more authors

    Engineering Optimization, 2025

    Abs DOI PDF Bib

    Predicting the physical properties of rubber presents difficulties because of the intricate and nonlinear relationships between its constituent elements. The laborious nature of rubber manufacture results in restricted data sets, hence exacerbating this difficulty. To achieve successful training of a neural network with a smaller amount of data, it is crucial to use an appropriate weight initialization approach and carefully construct the network architecture to avoid overfitting. This study presents a novel approach to representation learning that uses a contrastive loss with dynamic margin loss function. The approach is specifically designed to capture semantic similarities in rubber material quantities and physical-property recipes. In addition, a novel neural network architecture is proposed that uses latent vectors for assigning initial weights. The proposed approach, validated using data from a rubber manufacturing company, showed a significant improvement of up to 30% in predictive accuracy for physical properties with a restricted amount of data.

    @article{park2025contrastive,
      doi = {https://doi.org/10.1080/0305215X.2024.2346930},
      keywords = {Contrastive learning, Rubber physical properties, Dynamic margin, Representation learning, Deep learning, Manufacturing},
      title = {Contrastive loss with dynamic margin for rubber-recipe physical-property prediction},
      author = {Park, Hanbyeol and Kim, Kihun and Park, Kikun and Kwon, Jaeeun and Kang, Eungjun and Bae, Hyerim},
      journal = {Engineering Optimization},
      volume = {57},
      number = {4},
      pages = {1087--1110},
      year = {2025},
      publisher = {Taylor \& Francis}
    }

    Keywords: Contrastive learning Rubber physical properties Dynamic margin Representation learning Deep learning Manufacturing
21. 한국SCM학회지

    준지도 학습을 이용한 AIS 정박 패턴 분석과 비공식 묘박지 탐지 연구

    Jongmin Kim, Kikun Park, Hyunha Lee, and 2 more authors

    한국SCM학회지, 2025

    Abs DOI HTML Bib

    For efficient port logistics and precise maritime traffic management, automatic recognition of vessel anchoring behavior is essential. This study proposes a semi-supervised learning–based anchorage detection framework using AIS data to overcome the limitations of official anchorage information. By quantifying vessel stopping trajectory features and applying a stepwise semi-supervised structure based on high-confidence data, the method reduces labeling costs while maintaining generalization across ports. Trained with Busan and Ulsan data and validated with Yeosu–Gwangyang and Pyeongtaek–Dangjin data, the model achieved 98% accuracy and an F1-score of 0.98. The results demonstrate the potential of Automatic Identification System (AIS)-based automation for port activity recognition and its applicability to port operation optimization and real-time vessel monitoring.

    @article{kim2025ais,
      title = {준지도 학습을 이용한 AIS 정박 패턴 분석과 비공식 묘박지 탐지 연구},
      author = {Kim, Jongmin and Park, Kikun and Lee, Hyunha and Park, Hanbyeol and Bae, Hyerim},
      journal = {한국SCM학회지},
      doi = {10.25052/KSCM.2025.12.25.3.35},
      year = {2025},
      volume = {3},
      number = {특집호},
      pages = {35-47},
      keywords = {Automatic Identification System (AIS), Port and Anchorage Classification and Generalization, Semi-supervised Learning},
    }

    Keywords: Automatic Identification System (AIS) Port and Anchorage Classification and Generalization Semi-supervised Learning
22. 대한산업공학회

    작업 부하 기반 동적 충전 최적화 전략을 통한 AGV 운영 효율 개선

    Takhyeong Kim, Junhyuk Seo, Dohee Kim, and 2 more authors

    대한산업공학회 추계학술대회 논문집, 2025

    Bib

    @article{김탁형2025작업,
      title = {작업 부하 기반 동적 충전 최적화 전략을 통한 AGV 운영 효율 개선},
      author = {Kim, Takhyeong and Seo, Junhyuk and Kim, Dohee and Lee, Ho-in and Bae, Hyerim},
      journal = {대한산업공학회 추계학술대회 논문집},
      pages = {1845--1856},
      year = {2025},
      keywords = {AGV, Automated Guided Vehicle, Dynamic charging optimization, Workload-based scheduling, Operational efficiency}
    }

    Keywords: AGV Automated Guided Vehicle Dynamic charging optimization Workload-based scheduling Operational efficiency
23. 대한산업공학회

    해양 교통 밀집도 기반 선박 충돌 Near-miss 사고 식별 및 위험도 예측

    Somyeong Kim, Dohee Kim, and Hyerim Bae

    In , 2025

    HTML Bib

    @inproceedings{김소명2025Nearmiss,
      title = {해양 교통 밀집도 기반 선박 충돌 Near-miss 사고 식별 및 위험도 예측},
      author = {Kim, Somyeong and Kim, Dohee and Bae, Hyerim},
      year = {2025},
      pages = {1819 - 1830},
      journal = {대한산업공학회 추계학술대회 논문집},
      keywords = {Nearmiss,AIS data,Maritime collision,Traffic density,Machine learning}
    }

    Keywords: Nearmiss AIS data Maritime collision Traffic density Machine learning
24. 대한산업공학회지

    장비의 고장 단계 함수를 이용한 상태 지표 추출

    Jungho Choo, Hanbyeol Park, Yeonkyung Park, and 1 more author

    대한산업공학회지, 2025

    Abs DOI HTML Bib

    The Health Indicator (HI) is used to monitor equipment conditions. Supervised HI extraction methods assume a general degradation pattern, ensuring robustness. However, existing degradation functions are often designed assuming monotonic degradation without reflecting actual equipment characteristics. When the assumed pattern deviates from reality, the extracted HI may suffer from performance loss. This study aims to enhance HI extraction by identifying failure stages and designing degradation functions that better reflect real degradation patterns. A change point detection algorithm was used to identify failure stages, which were then used to construct the degradation function. The proposed method was validated using real-world data from a Weapon System. Compared to conventional linear and exponential degradation functions, the proposed approach improved HI performance by up to 29.13% in terms of Monotonicity. As the proposed degradation function can be applied to various equipment types, it is expected to enhance HI extraction in practical applications.

    @article{추정호2025장비,
      keywords = {Health Indicator, Degradation function, Change point detection, Condition-based maintenance, Weapon system},
      author = {Choo, Jungho and Park, Hanbyeol and Park, Yeonkyung and Bae, Hyerim},
      title = {장비의 고장 단계 함수를 이용한 상태 지표 추출},
      journal = {대한산업공학회지},
      doi = {10.7232/JKIIE.2025.51.5.412},
      issn = {1225-0988},
      year = {2025},
      volume = {51},
      number = {5},
      pages = {412-424}
    }

    Keywords: Health Indicator Degradation function Change point detection Condition-based maintenance Weapon system
25. 대한산업공학회지

    센서 데이터의 시계열 특성을 고려한 선박 연료 소모량 예측 프레임워크

    Hyunha Lee, Dohee Kim, Jungho Choo, and 2 more authors

    대한산업공학회지, 2025

    Abs DOI HTML PDF Bib

    The maritime industry plays a central role in global trade but also contributes significantly to environmental degradation due to high fuel consumption. Accordingly, the International Maritime Organization (IMO) has introduced environmental regulations, including the Carbon Intensity Indicator (CII). Complying with these regulations requires accurate fuel consumption forecasting to support optimal voyage planning. However, ship operation data frequently contain missing and anomalous values due to sensor failures, which reduce the accuracy of forecasts that rely on time-series characteristics. Although previous studies have applied machine learning and time-series models, they often fail to employ temporal dependencies or maintain data continuity. This study proposes a time-series forecasting framework that separates movement and anchorage states and performs imputation using environmental variables and similar trajectory group. The proposed method effectively leverages sensor data, despite frequent missing and anomalous values, and improves forecasting performance. The proposed framework facilitates voyage optimization by enabling accurate speed estimation and ensuring compliance with CII regulations.

    @article{ART003254703,
      keywords = {Ship fuel consumption, Time Series Forecasting, Missing data imputation, CII regulations, Maritime sustainability},
      author = {Lee, Hyunha and Kim, Dohee and Choo, Jungho and Jo, Sangmin and Bae, Hyerim},
      title = {센서 데이터의 시계열 특성을 고려한 선박 연료 소모량 예측 프레임워크},
      doi = {10.7232/JKIIE.2025.51.5.425},
      journal = {대한산업공학회지},
      issn = {1225-0988},
      year = {2025},
      volume = {51},
      number = {5},
      pages = {425-438}
    }

    Keywords: Ship fuel consumption Time Series Forecasting Missing data imputation CII regulations Maritime sustainability
26. 대한산업공학회지

    항만 혼잡 정보를 반영한 2-Stage 선박 도착 시간 예측 방법론 연구

    Seonghan Kim, Mingi Han, Minseop Kim, and 1 more author

    대한산업공학회지, 2025

    HTML Bib

    @article{김성한2025항만,
      title = {항만 혼잡 정보를 반영한 2-Stage 선박 도착 시간 예측 방법론 연구},
      author = {Kim, Seonghan and Han, Mingi and Kim, Minseop and Bae, Hyerim},
      keywords = {Vessel arrival time prediction, Port congestion, Two-stage prediction, Maritime logistics},
      journal = {대한산업공학회지},
      pages = {4369--4379},
      year = {2025},
    }

    Keywords: Vessel arrival time prediction Port congestion Two-stage prediction Maritime logistics
27. 인공위성 이미지 기반 미세먼지 예측 및 고해상도 공간 해석 프레임워크

    Eunhee Park, Minseop Kim, Hanbyeol Park, and 1 more author

    대한산업공학회지, 2025

    HTML Bib

    @article{박은희2025인공위성,
      title = {인공위성 이미지 기반 미세먼지 예측 및 고해상도 공간 해석 프레임워크},
      author = {Park, Eunhee and Kim, Minseop and Park, Hanbyeol and Bae, Hyerim},
      keywords = {Satellite image, Fine dust prediction, High-resolution spatial analysis, Deep learning, Environmental monitoring},
      journal = {대한산업공학회지},
      pages = {2268--2276},
      year = {2025},
    }

    Keywords: Satellite image Fine dust prediction High-resolution spatial analysis Deep learning Environmental monitoring
28. 대한산업공학회

    무기체계 상태기반정비를 위한 이미지 인코딩 기반 시계열 데이터 이상탐지

    Minseop Lee, Kikun Park, Hanbyeol Park, and 2 more authors

    대한산업공학회지, 2025

    Abs PDF Bib

    The radar detection weapon system is a surveillance and reconnaissance system designed to precisely detect long-range targets and determine their position and status. A critical component of the radar detection weapon system is the wave guide, which serves as a path for high-frequency signals. Maintaining appropriate temperature and humidity levels is essential to ensure seamless signal transmission, and this is regulated using an air compressor. However, if compressed air leakage occurs, temperature and humidity may rise, leading to performance degradation and potential component failure, necessitating early fault detection. The sensor data collected form actual operations exhibits high-dimensional and nonlinear characteristics, with abnormal patterns appearing irregularly, making numerical approaches challenging. Furthermore, such methods face limitations in effectively learning overall patterns and dependencies over time. In this paper, we propose an anomaly detection model that combines time-domain feature with image encoding to effectively capture nonlinear signal patterns. Experimental result using real-world operational data demonstrate that the proposed method outperforms existing approaches. The proposed approach is expected to be useful for applying Condition-Based Maintenance(CBM) to various equipment that utilizes time-series data.

    @article{이민섭2025무기체계,
      title = {무기체계 상태기반정비를 위한 이미지 인코딩 기반 시계열 데이터 이상탐지},
      author = {Lee, Minseop and Park, Kikun and Park, Hanbyeol and Park, Yeon-kyung and Bae, Hyerim},
      keywords = {Anomaly detection, Image encoding, Time Series Forecasting, Condition-based maintenance, Weapon system},
      journal = {대한산업공학회지},
      volume = {51},
      number = {4},
      pages = {276--287},
      year = {2025}
    }

    Keywords: Anomaly detection Image encoding Time Series Forecasting Condition-based maintenance Weapon system
29. 대한산업공학회

    LLM 기반의 시계열 예측 모형을 활용한 컨테이너 물동량 예측 방법론 연구

    Minseop Kim, Jaeeun Kwon, Taekhyun Park, and 2 more authors

    대한산업공학회 춘계공동학술대회 논문집, 2025

    HTML Bib

    @article{kim2025llm,
      title = {LLM 기반의 시계열 예측 모형을 활용한 컨테이너 물동량 예측 방법론 연구},
      author = {Kim, Minseop and Kwon, Jaeeun and Park, Taekhyun and Hong, Seongmoon and Bae, Hyerim},
      keywords = {Large Language Models, Container throughput prediction, Time Series Forecasting, Deep learning},
      journal = {대한산업공학회 춘계공동학술대회 논문집},
      pages = {2303--2312},
      year = {2025},
    }

    Keywords: Large Language Models Container throughput prediction Time Series Forecasting Deep learning
30. 대한산업공학회

    현실적 QC 작업 계획 수립을 위한 TSP 기반 프레임워크

    Minhee Kim, Kikun Park, Kihun Kim, and 1 more author

    대한산업공학회 춘계공동학술대회 논문집, 2025

    HTML Bib

    @article{김민희2025현실적,
      title = {현실적 QC 작업 계획 수립을 위한 TSP 기반 프레임워크},
      author = {Kim, Minhee and Park, Kikun and Kim, Kihun and Bae, Hyerim},
      keywords = {Quay crane scheduling, TSP, Container terminal, Work planning, Optimization},
      journal = {대한산업공학회 춘계공동학술대회 논문집},
      pages = {4342--4351},
      year = {2025}
    }

    Keywords: Quay crane scheduling TSP Container terminal Work planning Optimization
31. 한국전자거래학회지

    랜덤 포레스트와 전자 데이터 교환 정보를 활용한 수입 컨테이너 체류 시간 예측 연구

    Hanbyeol Park, Kikun Park, Jaehyeon Heo, and 3 more authors

    한국전자거래학회지, 2025

    Abs DOI PDF Bib

    연간 수백만 Twenty-foot Equivalent Unit (TEU) 컨테이너를 하역하는 항만에서, 제한된 야드 공간 내에서 효율적인 컨테이너 관리는 중요한 문제로 여겨진다. 하지만, 야드 운용을 효율적으로 관리하기에 많은 난항을 겪고 있는데, 그 원인 중 하나는 Import Container Dwell Time (ICDT)이 서로 다른 원인에서 기인한다. ICDT는 선박을 통해 양하 된 컨테이너를 운송사가 육지로 반출할 때까지 소요되는 시간으로 정의되는데, 컨테이너의 속성과 운송사의 배차계획 등 여러 복잡한 요인들에 영향을 받기 때문에, 정확히 예측하기 어렵다. 이러한 어려움에도 불구하고, 정확한 ICDT 예측은 효율적인 야드 운영에 기여하므로 매우 중요하다. 본 연구에서는 ICDT의 예측 성능을 향상시키기 위해 Electric Data Interchange (EDI) 데이터를 활용할 것을 제안한다. EDI는 컨테이너의 반입부터 반출까지의 프로세스 내 세관 신고, 컨테이너 반출 마감 기한과 같은 시간 정보를 포함하고 있다. 이러한 시간 정보들을 사용하여 ICDT 예측에 사용할 수 있는 외부 변수들을 생성할 수 있으며, 이는 ICDT 예측에 기여할 수 있다. 제안하는 접근법의 효과성을 검증하기 위해 실제 현장에서 수집된 데이터를 사용하였다. 실험 결과, 다양한 예측 모델 중 Random Forest (RF)가 가장 우수한 성능을 보여주었으며, EDI 데이터를 통합함으로써 예측 성능이 크게 개선됨을 확인하였다. 특히, EDI 데이터에서 파생된 변수들이 높은 변수 중요도를 보여주었다. 이러한 결과는 ICDT의 정확한 예측을 위해 EDI 정보의 필요성을 시사한다.

    @article{Park2025랜덤,
      title = {랜덤 포레스트와 전자 데이터 교환 정보를 활용한 수입 컨테이너 체류 시간 예측 연구},
      author = {Park, Hanbyeol and Park, Kikun and Heo, Jaehyeon and Kim, Minseop and Nam, Jaehyeon and Bae, Hyerim},
      keywords = {Container dwell time prediction, Random forest, EDI data, Port operations, Machine learning},
      journal = {한국전자거래학회지},
      volume = {30},
      number = {1},
      pages = {203--225},
      doi = {10.7838/jsebs.2025.30.1.203},
      year = {2025}
    }

    Keywords: Container dwell time prediction Random forest EDI data Port operations Machine learning

2024

1. arXiv

   A Surrogate Model for Quay Crane Scheduling Problem

   Kikun Park and Hyerim Bae

   2024

   Abs arXiv Bib

   In ports, a variety of tasks are carried out, and scheduling these tasks is crucial due to its significant impact on productivity, making the generation of precise plans essential. This study proposes a method to solve the Quay Crane Scheduling Problem (QCSP), a representative task scheduling problem in ports known to be NP-Hard, more quickly and accurately. First, the study suggests a method to create more accurate work plans for Quay Cranes (QCs) by learning from actual port data to accurately predict the working speed of QCs. Next, a Surrogate Model is proposed by combining a Machine Learning (ML) model with a Genetic Algorithm (GA), which is widely used to solve complex optimization problems, enabling faster and more precise exploration of solutions. Unlike methods that use fixed-dimensional chromosome encoding, the proposed methodology can provide solutions for encodings of various dimensions. To validate the performance of the newly proposed methodology, comparative experiments were conducted, demonstrating faster search speeds and improved fitness scores. The method proposed in this study can be applied not only to QCSP but also to various NP-Hard problems, and it opens up possibilities for the further development of advanced search algorithms by combining heuristic algorithms with ML models.

   @misc{park2024surrogate,
     title = {A Surrogate Model for Quay Crane Scheduling Problem},
     keywords = {Quay crane scheduling, Surrogate model, Genetic algorithm, Machine learning, Container terminal operations, Optimization},
     author = {Park, Kikun and Bae, Hyerim},
     journal = {arXiv},
     year = {2024}
   }

   Keywords: Quay crane scheduling Surrogate model Genetic algorithm Machine learning Container terminal operations Optimization
2. MARTRA

   Identifying key factors influencing import container dwell time using eXplainable Artificial Intelligence

   Yongjae Lee, Kikun Park, Hyunjae Lee, and 3 more authors

   Maritime Transport Research, 2024

   Abs DOI HTML Bib

   

   In a container terminal, the length of time that containers remain in the yard, known as Container Dwell Time (CDT), is considered one of the significant operational indicators due to its direct correlation with terminal productivity and efficiency. However, due to complex processing procedure and the involvement of various logistics stakeholders, CDT is subject to high uncertainty, making it more difficult for the terminal to manage. To address this issue, this paper presents a comprehensive framework to identify the Key Factors (KFs) influencing prolongation of CDT for import containers. In order to elucidate abnormal cases from dataset which contains yard loading information, the Process Mining (PM) method is utilized. Subsequently, XAI has been utilized to identify the KFs of import CDT. To reflect reality as closely as possible, we collected event data from a container terminal in Busan, Korea. Based on experiments, the KFs thus identified were: 1) Temperature, 2) Weight of container, 3) Voyage number of container 4) Block, 5) Shipping company, and 6) Month of discharging. To conclude, we formulated domain knowledge-based interpretations of the six most influential KFs.

   @article{lee2024identifying,
     title = {Identifying key factors influencing import container dwell time using eXplainable Artificial Intelligence},
     keywords = {Container dwell time, Explainable AI, Process mining, Container terminal operations, Feature selection, SHAP},
     author = {Lee, Yongjae and Park, Kikun and Lee, Hyunjae and Son, Jongpyo and Kim, Seonhwan and Bae, Hyerim},
     journal = {Maritime Transport Research},
     doi = {https://doi.org/10.1016/j.martra.2024.100116},
     volume = {7},
     pages = {100116},
     year = {2024},
     publisher = {Elsevier}
   }

   Keywords: Container dwell time Explainable AI Process mining Container terminal operations Feature selection SHAP
3. CogMI

   Neural Bezier Interpolation with Manifold Learning for Reliable Vessel Trajectory Prediction

   Gawon Lee, Dohee Kim, Sekil Park, and 3 more authors

   In 2024 IEEE 6th International Conference on Cognitive Machine Intelligence (CogMI), 2024

   PDF Bib

   @inproceedings{lee2024neural,
     title = {Neural Bezier Interpolation with Manifold Learning for Reliable Vessel Trajectory Prediction},
     author = {Lee, Gawon and Kim, Dohee and Park, Sekil and Sim, Sunghyun and Liu, Ling and Bae, Hyerim},
     keywords = {Vessel trajectory prediction, Bezier interpolation, Manifold learning, Deep learning, Maritime navigation},
     booktitle = {2024 IEEE 6th International Conference on Cognitive Machine Intelligence (CogMI)},
     pages = {189--196},
     year = {2024},
     organization = {IEEE}
   }

   Keywords: Vessel trajectory prediction Bezier interpolation Manifold learning Deep learning Maritime navigation
4. ICIC-B

   SIMULATION LEARNING FOR PREDICTING THE MAKESPAN FOR SCHEDULING PROBLEM

   Gawon Lee, Kikun Park, Muhammad Hanif Ramadhan, and 2 more authors

   ICIC Express Letters Part B : Applications, 2024

   Abs DOI PDF Bib

   Simulation models offer the advantage of unrestricted exploration of various solutions, including those for complex or difficult-to-formulate problems. In this context, simulation has been widely employed to address scheduling problems, which often demand extensive trial-and-error approaches. However, the computational time required to identify optimal solutions increases exponentially with problem size, rendering exhaustive evaluation of all possible scenarios impractical. Moreover, the prolonged search time makes the simulation method susceptible to disruptions such as unexpected machine failures or additional orders. To enhance the computational speed, we propose the application of Machine Learning (ML) models as an alternative to simulation. By leveraging ML models, we aim to preserve the expressive power of complex simulations while significantly reducing the required trial time. We evaluate the performance of various ML models and compare their prediction accuracy and computational efficiency. The results demonstrate promising prediction accuracy alongside a substantial reduction in computational time

   @article{leesimulation,
     title = {SIMULATION LEARNING FOR PREDICTING THE MAKESPAN FOR SCHEDULING PROBLEM},
     doi = {https://doi.org/10.24507/icicelb.15.10.1033},
     keywords = {Simulation learning, Makespan prediction, Scheduling problem, Machine learning, Computational efficiency},
     author = {Lee, Gawon and Park, Kikun and Ramadhan, Muhammad Hanif and Lee, Jungho and Bae, Hyerim},
     journal = {ICIC Express Letters Part B : Applications},
     year = {2024},
     publisher = {ICIC International}
   }

   Keywords: Simulation learning Makespan prediction Scheduling problem Machine learning Computational efficiency
5. ICIC-B

   Predictive Process Monitoring for Remaining Time Prediction with Transfer Learning

   Alif Nur Iman, Imam Mustafa Kamal, Muhammad Hanif Ramadhan, and 3 more authors

   革新的コンピューティング・情報・制御に関する速報, 2024

   Abs DOI PDF Bib

   Recent advancements in predictive process monitoring (PPM) have led to an improved performance by incorporating deep learning methodologies. However, some of these approaches employ a massive parameter architecture that requires a large dataset for training, posing challenges in many business process management scenarios. This study presents a transformer-based model with transfer learning for estimating remaining time prediction. We aim to improve the efficiency of business process management by accurately predicting the remaining time of a process. Additionally, we assessed the feasibility and transferability of the transfer learning method to enhance the performance of predictive process monitoring. We conducted experiments on four publicly available event logs. Our proposed architecture outperforms prior work, with a significant 53% improvement over the best baseline in the Helpdesk dataset. Additionally, the use of transfer learning leads to both positive and negative performance outcomes, depending on characteristics of the source and target process model.

   @article{nur2024predictive,
     title = {Predictive Process Monitoring for Remaining Time Prediction with Transfer Learning},
     author = {Iman, Alif Nur and Kamal, Imam Mustafa and Ramadhan, Muhammad Hanif and Kim, Dohee and Lee, Yongjae and Bae, Hyerim},
     keywords = {Predictive process monitoring, Transfer learning, Remaining time prediction, Transformer, Process mining},
     journal = {革新的コンピューティング・情報・制御に関する速報},
     doi = {10.24507/icicel.18.08.851},
     volume = {18},
     number = {08},
     pages = {851},
     year = {2024},
     publisher = {ICIC International 学会}
   }

   Keywords: Predictive process monitoring Transfer learning Remaining time prediction Transformer Process mining
6. IEEE

   A predictive discrete event simulation for predicting operation times in container terminal

   Kikun Park, Minseop Kim, and Hyerim Bae

   IEEE Access, 2024

   Abs DOI HTML PDF Bib

   Container terminals (CTs) play a crucial role within the global supply chain in the context of the global container transportation business. The primary obligation associated with these terminals involves ensuring the punctual execution of primary vessel operations, while strictly adhering to the Estimated Time of Departure (ETD) for vessels. To accomplish this goal, it is crucial to develop rigorous and exact operating strategies. Nonetheless, the accurate prediction of operation times in CT presents a significant challenge due to the concurrent involvement of various Container Handling Equipment (CHE) and the occurrence of unforeseen situations. To address this issue, this study proposes a novel approach called Predictive Discrete Event Simulation (PDES) that utilizes data collected from CT to predict the operation times. The PDES is an advanced approach that builds upon the widely used Discrete Event Simulation (DES) technique in the field of simulations. It is specifically designed to provide precise predictions of the operation times in CTs, where multiple events take place concurrently. The PDES in this paper aims to overcome the shortcomings of current simulation-based approaches for prediction in CT. These approaches often rely on predefined task sequences and assumed time for job handling time of CHE in their scenarios, which can result in reduced accuracy when predicting operation times. Through the resolution of these issues, the proposed PDES exhibits the capacity to improve predictive performance. To enhance the predictive performance of operation times in CTs through PDES, two approaches are introduced. The first approach entails the application of Support Vector Machine (SVM) algorithms for the purpose of predicting operation times. This approach is further augmented by integrating it with DES to improve the accuracy of predictive performance. The second approach involves predicting by simulating real-world operational scenarios in CTs using algorithms for CHE assignment. The predictive performance of the proposed PDES is assessed through the utilization of data gathered from Busan Port Terminal (BPT) in South Korea, demonstrating superior performance compared to alternative prediction approaches.

   @article{park2024predictive,
     keywords = {Container terminal operations, Discrete Event Simulation, Operation time prediction, Machine learning, Deep learning},
     title = {A predictive discrete event simulation for predicting operation times in container terminal},
     doi = {10.1109/ACCESS.2024.3389961},
     author = {Park, Kikun and Kim, Minseop and Bae, Hyerim},
     journal = {IEEE Access},
     volume = {12},
     pages = {58801--58822},
     year = {2024},
     publisher = {IEEE}
   }

   Keywords: Container terminal operations Discrete Event Simulation Operation time prediction Machine learning Deep learning
7. Adv.Eng.Info

   Semi-supervised binary classification with latent distance learning

   Imam Mustafa Kamal and Hyerim Bae

   Advanced Engineering Informatics. JCR Q1, Top 1.9%!!! , 2024

   Abs DOI arXiv HTML Bib

   Binary classification (BC) is a practical task that is ubiquitous in real-world problems, such as distinguishing healthy and unhealthy objects in biomedical diagnostics and defective and non-defective products in manufacturing inspections. Nonetheless, fully annotated data are commonly required to effectively solve this problem, and their collection by domain experts is a tedious and expensive procedure. In contrast to BC, several significant semi-supervised learning techniques that heavily rely on stochastic data augmentation techniques have been devised for solving multi-class classification. In this study, we demonstrate that the stochastic data augmentation technique is less suitable for solving typical BC problems because it can omit crucial features that strictly distinguish between positive and negative samples. To address this issue, we propose a new learning representation to solve the BC problem using a few labels with a random pair cross-distance learning mechanism. First, by harnessing a few labeled samples, the encoder network learns the projection of positive and negative samples in angular spaces to maximize and minimize their inter-class and intra-class distances, respectively. Second, the classifier learns to discriminate between positive and negative samples using on-the-fly labels generated based on the angular space and labeled samples to solve BC tasks. Extensive experiments were conducted using four real-world publicly available BC datasets. With few labels and without any data augmentation techniques, the proposed method outperformed state-of-the-art semi-supervised and self-supervised learning methods. Moreover, with 10% labeling, our semi-supervised classifier could obtain competitive accuracy compared with a fully supervised setting.

   @article{kamal2024semi,
     doi = {https://doi.org/10.1016/j.aei.2024.102441},
     keywords = {Semi-supervised learning, Binary classification, Latent distance learning, Representation learning, Angular space},
     title = {Semi-supervised binary classification with latent distance learning},
     author = {Kamal, Imam Mustafa and Bae, Hyerim},
     journal = {Advanced Engineering Informatics},
     volume = {61},
     pages = {102441},
     year = {2024},
     publisher = {Elsevier}
   }

   Keywords: Semi-supervised learning Binary classification Latent distance learning Representation learning Angular space
8. ICIC-B

   A Robust Explainable Feature Selection Method Based on Order Statistics

   Dohee Kim, Sangjae Lee, Sunghyun Sim, and 1 more author

   革新的コンピューティング・情報・制御に関する速報, 2024

   Abs DOI PDF Bib

   Recently, Feature Selection (FS) methods have been extensively researched using eXplainable Artificial Intelligence (XAI). Among these methods, SHapley Additive exPlanations (SHAP) is a representative approach. SHAP evaluates the impact of feature subset on predicted values based on game theory. Consequently, the feature importance can vary when iterated, and change depending on the prediction model used. In this paper, we propose a robust FS method based on order statistics. We determine feature ranking through two approaches: Feature Importance (FI) derived from the model-specific using different prediction models and model-agnostic using iterative experiments. Finally, we construct feature subsets based on the sum of these rankings and criteria. Through experiments, we validate the robustness and efficiency of our approach. By utilizing this approach, we can identify suitable feature subsets without the need to explore various FS methodologies, regardless of the prediction model and data dimension.

   @article{dohee2024robust,
     keywords = {Feature selection, Explainable AI, SHAP, Order statistics, Machine learning},
     doi = {10.24507/icicel.18.03.255},
     title = {A Robust Explainable Feature Selection Method Based on Order Statistics},
     author = {Kim, Dohee and Lee, Sangjae and Sim, Sunghyun and Bae, Hyerim},
     journal = {革新的コンピューティング・情報・制御に関する速報},
     volume = {18},
     number = {03},
     pages = {255},
     year = {2024},
     publisher = {ICIC International 学会}
   }

   Keywords: Feature selection Explainable AI SHAP Order statistics Machine learning
9. 대한산업공학회

   딥러닝 기반 직접 발화 음성과 딥보이스 판별 시스템 구축

   Kyunghoon Lee, Sujin Cho, Kikun Park, and 1 more author

   대한산업공학회 추계학술대회 논문집, 2024

   Bib

   @article{이경훈2024딥러닝,
     title = {딥러닝 기반 직접 발화 음성과 딥보이스 판별 시스템 구축},
     author = {Lee, Kyunghoon and Cho, Sujin and Park, Kikun and Bae, Hyerim},
     keywords = {Deep learning, Voice recognition, Deepfake detection, Speech classification, Audio analysis},
     journal = {대한산업공학회 추계학술대회 논문집},
     pages = {418--428},
     year = {2024}
   }

   Keywords: Deep learning Voice recognition Deepfake detection Speech classification Audio analysis
10. 한국빅데이터학회지

    SCFI 지수 예측 모형의 성능 향상을 위한 텍스트 데이터 기반 문자열 가중치 추출 연구

    Minseop Kim, Jaehyeon Heo, Dohee Kim, and 1 more author

    한국빅데이터학회지, 2024

    Abs DOI HTML Bib

    다양한 매체를 통해 생성되는 비정형 데이터는 많은 정보를 내포하고 있으며, 이에 대한 활용의 필요성이 증가하고 있다. 하지만 텍스트 데이터를 효과적으로 정량화하여 예측 모델의 성능을 향상시키는 연구는 많이 이루어지고 있지 않다. 이러한 문제를 해결하기 위해 텍스트 데이터에 포함된 영향도를 정량화하여 외부 변수로 활용할 수 있는 Noun-Frequency Link Frequency 방법론을 제안한다. 제안하는 방법론은 텍스트 데이터를 활용해 주간 키워드와 키워드의 상대적 영향도를 정량화된 수치의 형태로 추출하며, 이러한 정보를 반영하여, 상하이 컨테이너 운임 지수(SCFI) 예측 모델의 성능을 향상시키는 효과적인 방법론이라는 것을 검증하였다. 본 연구를 통해, 사회적, 심리적 영향을 미치는 이벤트가 텍스트 데이터에 등장하였을 때, 이러한 정보를 포착하고 예측 성능을 향상시킴과 동시에 등장하는 키워드들을 분석하여 해석 가능성을 높일 수 있다.

    @article{김민섭2024scfi,
      doi = {10.36498/kbigdt.2024.9.2.123},
      keywords = {SCFI prediction, Text data, NF-LF methodology, Maritime freight index, Natural language processing},
      title = {SCFI 지수 예측 모형의 성능 향상을 위한 텍스트 데이터 기반 문자열 가중치 추출 연구},
      author = {Kim, Minseop and Heo, Jaehyeon and Kim, Dohee and Bae, Hyerim},
      journal = {한국빅데이터학회지},
      volume = {9},
      number = {2},
      pages = {123--134},
      year = {2024},
      publisher = {(사) 한국빅데이터학회}
    }

    Keywords: SCFI prediction Text data NF-LF methodology Maritime freight index Natural language processing
11. 대한산업공학회

    유전 알고리즘 기반의 지능형 네트워크 구조 탐색을 활용한 Automatic Identification System 데이터 복원 연구

    Jaehyeon Heo, Kikun Park, Hanbyeol Park, and 3 more authors

    대한산업공학회 추계학술대회 논문집, 2024

    Bib

    @article{허재현2024유전,
      title = {유전 알고리즘 기반의 지능형 네트워크 구조 탐색을 활용한 Automatic Identification System 데이터 복원 연구},
      author = {Heo, Jaehyeon and Park, Kikun and Park, Hanbyeol and Kim, Minseop and Park, Se-Gil and Bae, Hyerim},
      keywords = {Genetic algorithm, Neural architecture search, AIS Data, Deep learning, Maritime data},
      journal = {대한산업공학회 추계학술대회 논문집},
      pages = {1304--1313},
      year = {2024}
    }

    Keywords: Genetic algorithm Neural architecture search AIS Data Deep learning Maritime data
12. 한국전자거래학회지

    해양사고 데이터 기반 해역 그리드화 및 심각도 지표 도출 연구

    Dohee Kim, Sangjae Lee, Suhyeon Jo, and 2 more authors

    The Journal of Society for e-Business Studies, 2024

    Abs DOI HTML Bib

    우리나라는 해상 물동량 및 통항량이 증가하면서 해상교통 환경의 복잡도와 해양사고의 위험이 증가하고 있다. 해양사고는 다른 운송 수단에 비해 사고 비율은 상대적으로 낮지만 상당한 손실을 초래하는 심각한 사고로 이어지기도 한다. 해양사고 기록은 해양사고의 심판에 붙일 경우 해양안전심판원의 재결서를 통해 기록되고, 심판불필요처분에 대해서는 해양사고 통계에 기록되며 공공데이터로 공개되어 있다. 해양사고를 줄이기 위해 많은 연구들이 이루어졌다. 해양사고의 주요 원인을 규명하고, 심각도를 정의 및 도출하는 연구가 주로 이루어졌는데, 이들 연구는 주로 심판재결서와 해양사고통계 중 하나의 데이터만을 이용하였다. 본 연구에서는 해양사고의 원인을 도출하고, 심각도를 산출하는 데 있어 중요한 두 가지 데이터를 함께 고려할 수 있는 프레임워크를 제안한다. 해양사고에 대한 다각적 고려를 위해 심각도 기준을 재정의 하였으며, KeyBERT모델을 활용하여 사고의 주요 원인이 될 수 있는 키워드를 도출하였다. 또한 이를 해역별 그리드로 표현하여 사고의 심각도와 원인 키워드에 대해 시각화하여 정보를 제공한다. 추후 이를 해사 안전의 정량적 평가를 위한 안전지표 개발의 기초 자료로서 활용하고자 한다.

    @article{김도희2024해양사고,
      keywords = {Maritime Accident, Grid-based analysis, Severity index, KeyBERT, Maritime safety},
      doi = {10.7838/jsebs.2024.29.1.057},
      title = {해양사고 데이터 기반 해역 그리드화 및 심각도 지표 도출 연구},
      author = {Kim, Dohee and Lee, Sangjae and Jo, Suhyeon and Park, Sekil and Bae, Hyerim},
      journal = {The Journal of Society for e-Business Studies},
      volume = {29},
      number = {1},
      pages = {57--72},
      year = {2024}
    }

    Keywords: Maritime Accident Grid-based analysis Severity index KeyBERT Maritime safety
13. 대한산업공학회

    시공간정보를 활용한 선박 연료 소모량 예측 프레임워크

    Hyunha Lee, Jungho Choo, Sangmin Jo, and 4 more authors

    대한산업공학회 추계학술대회 논문집, 2024

    Bib

    @article{이현하2024시공간정보를,
      title = {시공간정보를 활용한 선박 연료 소모량 예측 프레임워크},
      author = {Lee, Hyunha and Choo, Jungho and Jo, Sangmin and Lee, Yongjae and Kim, Dohee and Bae, Hyerim and Kang, Seongpil},
      keywords = {Spatiotemporal data, Ship fuel consumption prediction, Maritime logistics, Deep learning},
      journal = {대한산업공학회 추계학술대회 논문집},
      pages = {1314--1324},
      year = {2024}
    }

    Keywords: Spatiotemporal data Ship fuel consumption prediction Maritime logistics Deep learning
14. 한국전자거래학회지

    상태기반 정비를 위한 초기 성능저하 시점 (First Prediction Time, FPT) 추정 연구

    Mingyu Park, Hanbyeol Park, Minseop Kim, and 3 more authors

    한국전자거래학회지, 2024

    Abs DOI Bib

    장비의 고장 및 성능의 저하를 방지하기 위해서는 시스템의 상태를 모니터링 하고 장비의 고장을 예측하는 것이 필요하다. 이는 상태기반정비(Condition Based Maintenance +, CBM+)라고 불리며, 이를 위해서는 기계 혹은 장비의 상태를 실시간으로 진단하고, 초기 고장 감지 및 열화 상태를 예측하는 예측유지관리(Prognostics and Health Management, PHM)가 필수적이다. 본 논문에서는 무기체계의 냉매 누출의 문제에서 CBM+를 구현하기 위한 초기 냉매 누출시점(First Prediction Time, FPT) 추정방법론을 제시한다. 이는 냉매 누출 상태인 건강상태지표(Health Indicator, HI)를 보다 효과적으로 표현하여, 잔존수명(Remaining Useful Lifetime, RUL) 예측 성능에 큰 기여를 할 수 있다. 제안하는 방법론의 효율성은 현재 운용중인 무기체계에서 수집된 데이터를 통해 검증되었으며, FPT을 반영한 RUL 예측 성능이 FPT를 반영하지 않은 경우에 비해 우수한 성능을 보였다. 본 연구 방법론은 유사한 특성을 가진 하위부품에 대한 CBM+ 적용에 유용한 가이드라인으로 활용될 수 있을 것으로 보인다.

    @article{박민규2024상태기반,
      doi = {10.7838/jsebs.2024.29.1.093},
      title = {상태기반 정비를 위한 초기 성능저하 시점 (First Prediction Time, FPT) 추정 연구},
      author = {Park, Mingyu and Park, Hanbyeol and Kim, Minseop and Kim, Dohee and Park, Yeonkyung and Bae, Hyerim},
      journal = {한국전자거래학회지},
      volume = {29},
      number = {1},
      pages = {93--114},
      year = {2024}
    }

15. arXiv

    Artificial Intelligence-based Smart Port Logistics Metaverse for Enhancing Productivity, Environment, and Safety in Port Logistics: A Case Study of Busan Port

    Sunghyun Sim, Dohee Kim, Kikun Park, and 1 more author

    2024

    Abs arXiv Bib

    The increase in global trade, the impact of COVID-19, and the tightening of environmental and safety regulations have brought significant changes to the maritime transportation market. The port logistics industry is under pressure to respond to these changes, and is increasingly adopting data-driven approaches to meet this need. Recently, the development of digital twins and the metaverse utilizing artificial intelligence and big data has been attracting significant attention in smart port research. In this paper, we review the main challenges of port logistics, which involve productivity improvement and environmental conservation and safety and risk management, and present the metaverse and digital twin frameworks that incorporate artificial intelligence-based methodologies to support decision-making regarding each challenge. Furthermore, we provide a comprehensive review of the current development status and future outlook of Busan Port-related data platforms and AI-based port logistics research, focusing on the challenges discussed, with the aim of contributing to the development of smart ports in Busan.

    @misc{sim2024metaverse,
      title = {Artificial Intelligence-based Smart Port Logistics Metaverse for Enhancing Productivity, Environment, and Safety in Port Logistics: A Case Study of Busan Port},
      author = {Sim, Sunghyun and Kim, Dohee and Park, Kikun and Bae, Hyerim},
      keywords = {Smart port, Metaverse, Artificial intelligence, Port logistics, Digital twin},
      journal = {arXiv preprint arXiv:2409.10519},
      year = {2024}
    }

    Keywords: Smart port Metaverse Artificial intelligence Port logistics Digital twin

2023

1. TPAMI

   Correlation recurrent units: A novel neural architecture for improving the predictive performance of time-series data

   Sunghyun Sim, Dohee Kim, and Hyerim Bae

   IEEE Transactions on Pattern Analysis and Machine Intelligence. JCR Q1, Top 0.85%!!!! , 2023

   Abs DOI arXiv HTML PDF Bib

   Time-series forecasting (TSF) is a traditional problem in the field of artificial intelligence, and models such as recurrent neural network, long short-term memory, and gate recurrent units have contributed to improving its predictive accuracy. Furthermore, model structures have been proposed to combine time-series decomposition methods such as seasonal-trend decomposition using LOESS. However, this approach is learned in an independent model for each component, and therefore, it cannot learn the relationships between the time-series components. In this study, we propose a new neural architecture called a correlation recurrent unit (CRU) that can perform time-series decomposition within a neural cell and learn correlations (autocorrelation and correlation) between each decomposition component. The proposed neural architecture was evaluated through comparative experiments with previous studies using four univariate and four multivariate time-series datasets. The results showed that long- and short-term predictive performance was improved by more than 10%. The experimental results indicate that the proposed CRU is an excellent method for TSF problems compared to other neural architectures.

   @article{sim2023correlation,
     keywords = {Time Series Forecasting, Correlation recurrent unit, Time series decomposition, Neural Network, Autocorrelation, Deep learning},
     title = {Correlation recurrent units: A novel neural architecture for improving the predictive performance of time-series data},
     author = {Sim, Sunghyun and Kim, Dohee and Bae, Hyerim},
     journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
     volume = {45},
     number = {12},
     pages = {14266--14283},
     google_sholar_id = {roLk4NBRz8UC},
     year = {2023},
     publisher = {IEEE},
     doi = {10.1109/TPAMI.2023.3319557},
   }

   Keywords: Time Series Forecasting Correlation recurrent unit Time series decomposition Neural Network Autocorrelation Deep learning
2. Mar Sci&Eng

   Solving the inter-terminal truck routing problem for delay minimization using simulated annealing with normalized exploration rate

   Muhammad Hanif Ramadhan, Imam Mustafa Kamal, Dohee Kim, and 1 more author

   Journal of Marine Science and Engineering, 2023

   PDF Bib

   @article{ramadhan2023solving,
     title = {Solving the inter-terminal truck routing problem for delay minimization using simulated annealing with normalized exploration rate},
     author = {Ramadhan, Muhammad Hanif and Kamal, Imam Mustafa and Kim, Dohee and Bae, Hyerim},
     keywords = {Inter-terminal truck routing, Simulated annealing, Delay minimization, Container terminal operations, Optimization},
     journal = {Journal of Marine Science and Engineering},
     volume = {11},
     number = {11},
     pages = {2103},
     year = {2023},
     publisher = {MDPI}
   }

   Keywords: Inter-terminal truck routing Simulated annealing Delay minimization Container terminal operations Optimization
3. ICIC-B

   Ensemble Method for Feature Selection in Time-Series Prediction Based on Image Similarity

   Hyunjae Lee, Dohee Kim, Sangjae Lee, and 3 more authors

   革新的コンピューティング・情報・制御に関する速報-B: 応用, 2023

   PDF Bib

   @article{hyunjae2023ensemble,
     title = {Ensemble Method for Feature Selection in Time-Series Prediction Based on Image Similarity},
     author = {Lee, Hyunjae and Kim, Dohee and Lee, Sangjae and Park, Hanbyeol and Bae, Hyerim and Choi, Keonwoo},
     keywords = {Ensemble method, Feature selection, Time Series Forecasting, Image similarity, Deep learning},
     journal = {革新的コンピューティング・情報・制御に関する速報-B: 応用},
     volume = {14},
     number = {07},
     pages = {709},
     year = {2023},
     publisher = {ICIC International 学会}
   }

   Keywords: Ensemble method Feature selection Time Series Forecasting Image similarity Deep learning
4. 지능정보연구

   화물의 특성 및 적재 공간을 고려한 LCL 화물 적재 알고리즘

   Daesan Park, Sangmin Jo, Dongyoon Park, and 3 more authors

   지능정보연구, 2023

   Abs HTML Bib

   다품종 소량 생산 품목의 수요 증가와 전자상거래 시장의 발달로 인하여 소량화물(Less than Container Load, LCL)의수출입 비중이 확대되고 있다. 이에 따라 국제물류주선업에서 소량화물을 다루는 업체의 비중이 늘어나고 있다. 화물 크기의 다양성과 이해당사자의 이익 구조가 맞물려, 컨테이너의 공간 효율성을 최대화할 수 있는 화물 적재 방법론이 중요 해지고 있다. 그러나, 화물 적재 계획을 미리 수립하여 컨테이너 조작장(Container Freight Station, CFS)에 전달하는 현 상황의 특성상, 산업현장에서 확인 가능한 변수들을 적재 계획에 반영할 수 없다는 한계가 있다. 따라서, 본 연구에서는 산업현장에서 적재 계획의 수정이 용이한 화물 적재 방법론을 제시한다. 화물의 특성에 따른 불용 공간과 컨테이너 내부 현황을 고려할 수 있도록 하여 산업 현장의 요구사항을 반영하고, 3차원 공간을 2차원 평면 레이어로 조작하여 적재 계획을 수립하여 시간 복잡도를 줄임으로써 산업현장의 활용성을 높였다. 본 연구에서 제시하는 방법론을 통해 화물 적재 계획 품질의 일관성을 높이고, 적재 계획 수립 자동화에 기여를 할 수 있다.

   @article{박대산2023화물의,
     title = {화물의 특성 및 적재 공간을 고려한 LCL 화물 적재 알고리즘},
     keywords = {LCL cargo loading, Container loading algorithm, Space optimization, Logistics, 3D bin packing},
     author = {Park, Daesan and Jo, Sangmin and Park, Dongyoon and Lee, Yongjae and Kim, Dohee and Bae, Hyerim},
     journal = {지능정보연구},
     volume = {29},
     number = {4},
     pages = {375--393},
     year = {2023}
   }

   Keywords: LCL cargo loading Container loading algorithm Space optimization Logistics 3D bin packing
5. 한국빅데이터논문지

   대리운전 시장의 지역별 수요 예측 모형의성능 향상을 위한 방법론 연구

   Minseop Kim, Kikun Park, Jaehyeon Heo, and 2 more authors

   한국빅데이터논문지 제, 2023

   Bib

   @article{김민섭2023대리운전,
     title = {대리운전 시장의 지역별 수요 예측 모형의성능 향상을 위한 방법론 연구},
     author = {Kim, Minseop and Park, Kikun and Heo, Jaehyeon and Kwon, Jaeeun and Bae, Hyerim},
     keywords = {Demand prediction, Designated driver service, Regional forecasting, Machine learning},
     journal = {한국빅데이터논문지 제},
     volume = {8},
     number = {1},
     year = {2023}
   }

   Keywords: Demand prediction Designated driver service Regional forecasting Machine learning
6. 해운물류연구

   설명가능한 인공지능을 활용한 컨테이너 운임 예측 모형 연구

   Dohee Kim, Hyerim Bae, and Sangjae Lee

   해운물류연구, 2023

   Abs HTML Bib

   컨테이너 운임은 대외 수출 비중이 높은 우리나라 산업의 물류비용을 결정하는 중요한 요인 중 하나이며, 산업 및 물가에 직접적인 영향을 미치는 중요한 경제 지표 중 하나이다. COVID-19 발생 이후 강한 충격에 급격히 상승한 운임에 대해 관련 영향 요인을 파악하고, 이를 정확히 예측하는 모델의 필요성이 대두되었다. 본 연구에서는 컨테이너 운임 지수 중 하나인 SCFI에 대한 정확한 예측을 목표로 하는 딥러닝 모델 기반의 프레임워크를 제안하고, 설명할 수 있는 인공지능 방법론 중 하나인 SHAP을 이용하여 사후 분석을 수행하였다. GRU, LSTM, RNN 모형 중 GRU 모형이 가장 좋은 예측 성능을 보였으며, 1주, 4주 예측 시 90%이상의 높은 예측 정확도를 보였으나, 12주 예측 시 80%이상의 예측 정확도를 보였다. 또한 SHAP 분석을 통해 변수 별 중요도를 바탕으로 구성한 변수 조합으로 실험 시 예측 성능이 개선되는 것을 확인하였다. 이를 통해 딥러닝 방법론이 가지고 있던 고질적인 한계점인 설명력 문제를 일부 해소하고, 예측력 개선에서도 활용할 수 있을 것이라 기대된다.

   @article{김도희2023설명가능한,
     title = {설명가능한 인공지능을 활용한 컨테이너 운임 예측 모형 연구},
     keywords = {Container freight rate prediction, Explainable AI, SHAP, Deep learning, SCFI, Maritime economics},
     author = {Kim, Dohee and Bae, Hyerim and Lee, Sangjae},
     journal = {해운물류연구},
     volume = {39},
     number = {2},
     pages = {197--215},
     year = {2023}
   }

   Keywords: Container freight rate prediction Explainable AI SHAP Deep learning SCFI Maritime economics
7. 한국신뢰성학회

   무기체계 상태 기반 정비를 위한 인공신경망 기반 상태진단모델 개발

   Mingyu Park, Hanbyeol Park, Minseop Kim, and 4 more authors

   한국신뢰성학회 학술대회논문집, 2023

   Bib

   @article{박민규2023무기체계,
     title = {무기체계 상태 기반 정비를 위한 인공신경망 기반 상태진단모델 개발},
     author = {Park, Mingyu and Park, Hanbyeol and Kim, Minseop and Kwon, Jaeeun and Kim, Dohee and Bae, Hyerim and Park, Yeonkyung},
     journal = {한국신뢰성학회 학술대회논문집},
     pages = {60--60},
     year = {2023},
     keywords = {Weapon system, Condition-based maintenance, Neural Network, Condition diagnosis, Reliability engineering}
   }

   Keywords: Weapon system Condition-based maintenance Neural Network Condition diagnosis Reliability engineering
8. 대한산업공학회

   GPS 데이터 기반 내부트럭 경로 예측 모델 개발

   Sangjae Lee, Kikun Park, Alif Nur Iman, and 2 more authors

   대한산업공학회 춘계공동학술대회 논문집, 2023

   Bib

   @article{이상재2023gps,
     title = {GPS 데이터 기반 내부트럭 경로 예측 모델 개발},
     author = {Lee, Sangjae and Park, Kikun and Iman, Alif Nur and Ramadhan, Muhammad Hanif and Bae, Hyerim},
     journal = {대한산업공학회 춘계공동학술대회 논문집},
     pages = {4224--4233},
     year = {2023},
     keywords = {GPS data, Internal truck, Route prediction, Container terminal, Logistics optimization}
   }

   Keywords: GPS data Internal truck Route prediction Container terminal Logistics optimization
9. 대한산업공학회

   외부 충격도 반영을 위한 텍스트 데이터 기반 NF-LF 방법론 개발

   Jaehyeon Heo, Minseop Kim, Dohee Kim, and 2 more authors

   대한산업공학회 춘계공동학술대회 논문집, 2023

   Bib

   @article{허재현2023외부,
     title = {외부 충격도 반영을 위한 텍스트 데이터 기반 NF-LF 방법론 개발},
     author = {Heo, Jaehyeon and Kim, Minseop and Kim, Dohee and Park, Daesan and Bae, Hyerim},
     journal = {대한산업공학회 춘계공동학술대회 논문집},
     pages = {4213--4223},
     year = {2023},
     keywords = {External shock, Text data, NF-LF methodology, Time Series Forecasting, Natural language processing}
   }

   Keywords: External shock Text data NF-LF methodology Time Series Forecasting Natural language processing
10. 대한산업공학회

    항만 정시성 예측을 위한 선박 ETA/RTA 예측 알고리즘

    Suhyeon Jo, Kikun Park, Gawon Lee, and 2 more authors

    대한산업공학회 춘계공동학술대회 논문집, 2023

    Bib

    @article{조수현2023항만,
      title = {항만 정시성 예측을 위한 선박 ETA/RTA 예측 알고리즘},
      author = {Jo, Suhyeon and Park, Kikun and Lee, Gawon and Jo, Sangmin and Bae, Hyerim},
      journal = {대한산업공학회 춘계공동학술대회 논문집},
      pages = {4087--4095},
      year = {2023},
      keywords = {Port punctuality, ETA prediction, RTA prediction, Vessel arrival, Maritime logistics}
    }

    Keywords: Port punctuality ETA prediction RTA prediction Vessel arrival Maritime logistics
11. 대한산업공학회

    고무 물성 예측을 위한 클러스터링 기반 프레임워크 개발

    Jaeeun Kwon, Hanbyeol Park, Mingyu Park, and 2 more authors

    대한산업공학회 춘계공동학술대회 논문집, 2023

    Bib

    @article{권재은2023고무,
      title = {고무 물성 예측을 위한 클러스터링 기반 프레임워크 개발},
      author = {Kwon, Jaeeun and Park, Hanbyeol and Park, Mingyu and Jung, Hyokyeong and Bae, Hyerim},
      journal = {대한산업공학회 춘계공동학술대회 논문집},
      pages = {2512--2520},
      year = {2023},
      keywords = {Rubber property prediction, Clustering, Machine learning, Material science, Framework development}
    }

    Keywords: Rubber property prediction Clustering Machine learning Material science Framework development
12. 대한산업공학회

    독립변수들 간의 상호관계 학습을 위한 CoRepNet

    Sunghyun Sim, Dohee Kim, and Hyerim Bae

    대한산업공학회 춘계공동학술대회 논문집, 2023

    Bib

    @article{심성현2023독립변수들,
      title = {독립변수들 간의 상호관계 학습을 위한 CoRepNet},
      author = {Sim, Sunghyun and Kim, Dohee and Bae, Hyerim},
      journal = {대한산업공학회 춘계공동학술대회 논문집},
      pages = {2115--2127},
      year = {2023},
      keywords = {CoRepNet, Variable correlation, Deep learning, Feature interaction, Representation learning}
    }

    Keywords: CoRepNet Variable correlation Deep learning Feature interaction Representation learning
13. Trans. Serv. Comput.

    Metric learning as a service with covariance embedding

    Imam Mustafa Kamal, Hyerim Bae, and Ling Liu

    IEEE Transactions on Services Computing, 2023

    Bib

    @article{kamal2023metric,
      title = {Metric learning as a service with covariance embedding},
      author = {Kamal, Imam Mustafa and Bae, Hyerim and Liu, Ling},
      keywords = {Metric learning, Covariance embedding, Service computing, Deep learning, Representation learning},
      journal = {IEEE Transactions on Services Computing},
      volume = {16},
      number = {5},
      pages = {3508--3522},
      year = {2023},
      publisher = {IEEE}
    }

    Keywords: Metric learning Covariance embedding Service computing Deep learning Representation learning
14. 해양환경안전학회지

    Maritime Safety Tribunal Ruling Analysis using SentenceBERT

    Bori Yoon, Sekil Park, Hyerim Bae, and 1 more author

    Journal of the Korean Society of Marine Environment & Safety, 2023

    Bib

    @article{yoon2023maritime,
      title = {Maritime Safety Tribunal Ruling Analysis using SentenceBERT},
      author = {Yoon, Bori and Park, Sekil and Bae, Hyerim and Sim, Sunghyun},
      keywords = {Maritime safety, Tribunal ruling analysis, SentenceBERT, Natural language processing, Text analysis},
      journal = {Journal of the Korean Society of Marine Environment \& Safety},
      volume = {29},
      number = {7},
      pages = {843--856},
      year = {2023},
      publisher = {The Korean Society of Marine Environment and Safety}
    }

    Keywords: Maritime safety Tribunal ruling analysis SentenceBERT Natural language processing Text analysis
15. IEEE BigData

    A novel training mechanism for health indicator construction and remaining useful lifetime (RUL) prediction

    Hanbyeol Park, Dohee Kim, Minseop Kim, and 3 more authors

    In 2023 IEEE International Conference on Big Data and Smart Computing (BigComp), 2023

    PDF Bib

    @inproceedings{park2023novel,
      title = {A novel training mechanism for health indicator construction and remaining useful lifetime (RUL) prediction},
      author = {Park, Hanbyeol and Kim, Dohee and Kim, Minseop and Park, Myunggwan and Bae, Hyerim and Sim, Sunghyun},
      keywords = {Health indicator, Remaining Useful Life Prediction, Deep learning, Predictive maintenance, Training mechanism},
      booktitle = {2023 IEEE International Conference on Big Data and Smart Computing (BigComp)},
      pages = {379--382},
      year = {2023},
      organization = {IEEE}
    }

    Keywords: Health indicator Remaining Useful Life Prediction Deep learning Predictive maintenance Training mechanism
16. IEEE CogMI

    Improving time-series classification accuracy based on temporal feature representation learning using CRU-LSTM autoencoder

    Dohee Kim, Hanbyeol Park, Sunghyun Sim, and 3 more authors

    In 2023 IEEE 5th International Conference on Cognitive Machine Intelligence (CogMI), 2023

    PDF Bib

    @inproceedings{kim2023cruauto,
      title = {Improving time-series classification accuracy based on temporal feature representation learning using CRU-LSTM autoencoder},
      author = {Kim, Dohee and Park, Hanbyeol and Sim, Sunghyun and Yoon, Bori and Liu, Ling and Bae, Hyerim},
      keywords = {Time series classification, CRU-LSTM, Autoencoder, Feature representation, Deep learning},
      booktitle = {2023 IEEE 5th International Conference on Cognitive Machine Intelligence (CogMI)},
      pages = {98--105},
      year = {2023},
      organization = {IEEE}
    }

    Keywords: Time series classification CRU-LSTM Autoencoder Feature representation Deep learning

2022

1. IEEE

   A data-driven recipe simulation for synthetic rubber production

   Kikun Park, Hanbyeol Park, and Hyerim Bae

   IEEE Access, 2022

   PDF Bib

   @article{park2022data,
     title = {A data-driven recipe simulation for synthetic rubber production},
     author = {Park, Kikun and Park, Hanbyeol and Bae, Hyerim},
     keywords = {Data-driven simulation, Synthetic rubber production, Recipe optimization, Manufacturing, Machine learning},
     journal = {IEEE Access},
     volume = {10},
     pages = {129408--129418},
     year = {2022},
     publisher = {IEEE}
   }

   Keywords: Data-driven simulation Synthetic rubber production Recipe optimization Manufacturing Machine learning
2. TRD

   Deep collaborative learning model for port-air pollutants prediction using automatic identification system

   Sunghyun Sim, Jin-Hyoung Park, and Hyerim Bae

   Transportation Research Part D: Transport and Environment. JCR Q1, Top 8.62%!!! , 2022

   Abs HTML Bib

   Air pollution in port cities is aggravated by ship pollutant emissions. A deep collaborative learning (DCL)-based prediction model using automatic identification system (AIS) is proposed in this study to predict this type of pollution. In the model training process, a novel data preprocessing method was devised to efficiently handle heterogeneous data: air pollution, weather conditions, and AIS data. To combine these data together, a pretraining step is introduced using an autoencoder-based model, which is a customized convolutional long short-term memory network model, followed by a DCL method for the prediction of highly accurate air pollution values (for both short- and long-term predictions). Compared with other approaches, this method showed on an average, a performance improvement of nearly 10% in terms of the root mean squared error. Experiments to test and validate the model were conducted near the North/Old Busan Port, Republic of Korea.

   @article{sim2022deep,
     keywords = {Port air pollution, Deep learning, AIS data, LSTM, Ship Emissions, Environmental prediction},
     title = {Deep collaborative learning model for port-air pollutants prediction using automatic identification system},
     author = {Sim, Sunghyun and Park, Jin-Hyoung and Bae, Hyerim},
     journal = {Transportation Research Part D: Transport and Environment},
     volume = {111},
     pages = {103431},
     year = {2022},
     publisher = {Elsevier}
   }

   Keywords: Port air pollution Deep learning AIS data LSTM Ship Emissions Environmental prediction
3. PRL

   Cooperative auto-classifier networks for boosting discriminant capacity

   Imam Mustafa Kamal and Hyerim Bae

   Pattern Recognition Letters, 2022

   HTML Bib

   @article{kamal2022cooperative,
     title = {Cooperative auto-classifier networks for boosting discriminant capacity},
     author = {Kamal, Imam Mustafa and Bae, Hyerim},
     keywords = {Cooperative learning, Auto-classifier, Discriminant capacity, Deep learning, Pattern recognition},
     journal = {Pattern Recognition Letters},
     volume = {160},
     pages = {82--89},
     year = {2022},
     publisher = {Elsevier}
   }

   Keywords: Cooperative learning Auto-classifier Discriminant capacity Deep learning Pattern recognition
4. ICIC-B

   Improving the Performance of Mixing-Process Machine Classification Based on Feature Engineering Techniques: A Case Study in Rubber-Belt Industry

   Alif Nur Iman, Jinuk Kim, Jeong Su Son, and 2 more authors

   革新的コンピューティング・情報・制御に関する速報-B: 応用, 2022

   PDF Bib

   @article{nur2022improving,
     title = {Improving the Performance of Mixing-Process Machine Classification Based on Feature Engineering Techniques: A Case Study in Rubber-Belt Industry},
     author = {Iman, Alif Nur and Kim, Jinuk and Son, Jeong Su and Ramadhan, Muhammad Hanif and Bae, Hyerim},
     keywords = {Feature engineering, Machine classification, Rubber-belt industry, Mixing process, Manufacturing},
     journal = {革新的コンピューティング・情報・制御に関する速報-B: 応用},
     volume = {13},
     number = {06},
     pages = {639},
     year = {2022},
     publisher = {ICIC International 学会}
   }

   Keywords: Feature engineering Machine classification Rubber-belt industry Mixing process Manufacturing
5. Appl. Sci.

   Big data and AI for process innovation in the Industry 4.0 era

   Jaehun Park and Hyerim Bae

   Applied Sciences, 2022

   PDF Bib

   @article{park2022big,
     title = {Big data and AI for process innovation in the Industry 4.0 era},
     author = {Park, Jaehun and Bae, Hyerim},
     keywords = {Big data, Artificial intelligence, Industry 4.0, Process innovation, Smart manufacturing},
     journal = {Applied Sciences},
     volume = {12},
     number = {13},
     pages = {6346},
     year = {2022},
     publisher = {MDPI}
   }

   Keywords: Big data Artificial intelligence Industry 4.0 Process innovation Smart manufacturing
6. ICIC-B

   Obtaining lightweight human activity recognition model through knowledge distillation of deep neural network

   Muhammad Hanif Ramadhan, Taufik Nur Adi, Hyerim Bae, and 1 more author

   ICIC Express Letters, Part B: Applications, 2022

   PDF Bib

   @article{ramadhan2022obtaining,
     title = {Obtaining lightweight human activity recognition model through knowledge distillation of deep neural network},
     author = {Ramadhan, Muhammad Hanif and Adi, Taufik Nur and Bae, Hyerim and Kim, Hyemee},
     keywords = {Human activity recognition, Knowledge distillation, Deep learning, Lightweight model, Model compression},
     journal = {ICIC Express Letters, Part B: Applications},
     volume = {13},
     number = {5},
     pages = {519--526},
     year = {2022}
   }

   Keywords: Human activity recognition Knowledge distillation Deep learning Lightweight model Model compression
7. PR

   Super-encoder with cooperative autoencoder networks

   Imam Mustafa Kamal and Hyerim Bae

   Pattern Recognition. JCR Q1, Top 9.02%!!! , 2022

   Abs DOI HTML Bib

   Dimensionality reduction plays a crucial role in classification, object detection, and pattern recognition tasks. Its main objective is to decrease the dimension of the original data while retaining the most distinctive information. With the emergence of deep learning, an autoencoder has become a state-of-the-art non-linear dimensionality-reduction method. Nonetheless, as the existing autoencoder models are devised to follow the data distribution and employ similarity techniques, preserving distinctive information can be problematic. To tackle this issue, we propose super-encoder (SE) networks trained in a supervised and cooperative manner. The SE consists of an encoder, separator, and decoder networks. The encoder combined with separator networks are dedicated to generating separable latent representation based on the label, and the decoder network should be able to reconstruct it to the original data simultaneously. Herein, we introduce a novel cooperative learning mechanism with a new loss function; therefore, the encoder, separator, and decoder networks can cooperate to achieve these objectives. Extensive experiments using benchmark datasets were conducted. The results indicated that the SE is more effective in extracting separable latent code than the existing supervised and unsupervised dimensionality-reduction models. Furthermore, as a generator, it can obtain highly competitive realistic images.

   @article{kamal2022super,
     keywords = {Dimensionality reduction, Autoencoder, Super-encoder, Cooperative learning, Representation learning, Deep learning},
     doi = {https://doi.org/10.1016/j.patcog.2022.108562},
     title = {Super-encoder with cooperative autoencoder networks},
     author = {Kamal, Imam Mustafa and Bae, Hyerim},
     journal = {Pattern Recognition},
     volume = {126},
     pages = {108562},
     year = {2022},
     publisher = {Elsevier}
   }

   Keywords: Dimensionality reduction Autoencoder Super-encoder Cooperative learning Representation learning Deep learning
8. IEEE

   Automatic conversion of event data to event logs using CNN and event density embedding

   Sunghyun Sim, Riska Asriana Sutrisnowati, Seokrae Won, and 2 more authors

   IEEE Access, 2022

   PDF Bib

   @article{sim2022automatic,
     title = {Automatic conversion of event data to event logs using CNN and event density embedding},
     author = {Sim, Sunghyun and Sutrisnowati, Riska Asriana and Won, Seokrae and Lee, Sanghwa and Bae, Hyerim},
     keywords = {Event log conversion, CNN, Event density embedding, Process mining, Image encoding},
     journal = {IEEE Access},
     volume = {10},
     pages = {15994--16009},
     year = {2022},
     publisher = {IEEE}
   }

   Keywords: Event log conversion CNN Event density embedding Process mining Image encoding
9. 한국산업정보학회논문지

   벌크항만의 하역 최적화를 위한 정수계획모형

   Taesun Yu, Yusin Lee, Hyun-gon Park, and 2 more authors

   한국산업정보학회논문지, 2022

   Bib

   @article{유태선2022벌크항만의,
     title = {벌크항만의 하역 최적화를 위한 정수계획모형},
     author = {Yu, Taesun and Lee, Yusin and Park, Hyun-gon and Kim, Dohee and Bae, Hyerim},
     keywords = {Bulk port, Cargo handling optimization, Integer programming, Port operations, Mathematical modeling},
     journal = {한국산업정보학회논문지},
     volume = {27},
     number = {6},
     pages = {105--114},
     year = {2022},
     publisher = {한국산업정보학회}
   }

   Keywords: Bulk port Cargo handling optimization Integer programming Port operations Mathematical modeling
10. 한국빅데이터논문지

    LSTM-AutoEncoder 를 활용한선박 메인엔진의 이상 탐지 및 라벨링

    Dohee Kim, Yeongjae Han, Hyemee Kim, and 3 more authors

    한국빅데이터논문지 제, 2022

    Bib

    @article{김도희2022lstm,
      title = {LSTM-AutoEncoder 를 활용한선박 메인엔진의 이상 탐지 및 라벨링},
      author = {Kim, Dohee and Han, Yeongjae and Kim, Hyemee and Kang, Seongpil and Kim, Kihun and Bae, Hyerim},
      keywords = {LSTM, Ship main engine, Anomaly detection, Labeling, Predictive maintenance},
      journal = {한국빅데이터논문지 제},
      volume = {7},
      number = {1},
      year = {2022}
    }

    Keywords: LSTM Ship main engine Anomaly detection Labeling Predictive maintenance
11. 한국빅데이터학회

    밀도기반 선박 도메인을 이용한 안전 성능 지표 활용성 연구

    Yeongjae Han, Sunghyun Sim, and Hyerim Bae

    한국빅데이터학회지, 2022

    DOI PDF Bib

    @article{한영재2022밀도기반,
      doi = {10.36498/kbigdt.2022.7.1.89},
      title = {밀도기반 선박 도메인을 이용한 안전 성능 지표 활용성 연구},
      author = {Han, Yeongjae and Sim, Sunghyun and Bae, Hyerim},
      keywords = {Density-based ship domain, Safety performance indicator, Maritime safety, AIS Data},
      journal = {한국빅데이터학회지},
      volume = {7},
      number = {1},
      pages = {89--97},
      year = {2022},
      publisher = {(사) 한국빅데이터학회}
    }

    Keywords: Density-based ship domain Safety performance indicator Maritime safety AIS Data
12. IEEE Big Data

    TDTA: A new hybrid framework for long-term forecasting of container volume

    Dohee Kim and Hyerim Bae

    In 2022 IEEE International Conference on Big Data (Big Data), 2022

    PDF Bib

    @inproceedings{kim2022tdta,
      title = {TDTA: A new hybrid framework for long-term forecasting of container volume},
      author = {Kim, Dohee and Bae, Hyerim},
      keywords = {Container volume forecasting, Hybrid framework, Time Series Forecasting, Deep learning, Maritime logistics},
      booktitle = {2022 IEEE International Conference on Big Data (Big Data)},
      pages = {6718--6720},
      year = {2022},
      organization = {IEEE}
    }

    Keywords: Container volume forecasting Hybrid framework Time Series Forecasting Deep learning Maritime logistics
13. arXiv

    Automatic Discovery of Multi-perspective Process Model using Reinforcement Learning

    Sunghyun Sim, Ling Liu, and Hyerim Bae

    2022

    arXiv Bib

    @misc{sim2022autodiscovery,
      title = {Automatic Discovery of Multi-perspective Process Model using Reinforcement Learning},
      author = {Sim, Sunghyun and Liu, Ling and Bae, Hyerim},
      keywords = {Process mining, Reinforcement learning, Multi-perspective process model, Process discovery, Deep learning},
      journal = {arXiv preprint arXiv:2211.16687},
      year = {2022}
    }

    Keywords: Process mining Reinforcement learning Multi-perspective process model Process discovery Deep learning
14. ICIC-B

    Colored Petri Nets simulation on the cloud with python: A framework overview

    Muhammad Hanif Ramadhan, Imam Mustafa Kamal, Alif Nur Iman, and 3 more authors

    ICIC Express Letters, Part B: Applications, 2022

    PDF Bib

    @article{ramadhan2022cpn,
      title = {Colored Petri Nets simulation on the cloud with python: A framework overview},
      author = {Ramadhan, Muhammad Hanif and Kamal, Imam Mustafa and Iman, Alif Nur and Kim, Dohee and Bae, Hyerim and Park, You-Jin},
      keywords = {Colored Petri Nets, Cloud computing, Python, Simulation framework, Process modeling},
      journal = {ICIC Express Letters, Part B: Applications},
      volume = {13},
      number = {7},
      pages = {729--736},
      year = {2022}
    }

    Keywords: Colored Petri Nets Cloud computing Python Simulation framework Process modeling

2021

1. MARTRA

   Vessel estimated time of arrival prediction system based on a path-finding algorithm

   Kikun Park, Sunghyun Sim, and Hyerim Bae

   Maritime Transport Research, 2021

   Abs DOI HTML PDF Bib

   

   Port operation efficiency has grown in importance as container volumes and vessel sizes have increased. For improved port operations efficiency, the estimated time of arrival (ETA) of sea-going vessels must be accurately predicted. In this paper, an AIS data-driven method- ology is proposed for the estimation of vessel ETA at ports. For ETA prediction, we first introduce how to find possible vessel trajectories using AIS data mining methods and reinforcement learning (RL); next, we introduce the Markov Chain property and Bayesian Sampling to estimate the speed over ground (SOG) of a vessel. Experimentation comparing the proposed methodology with an existing one was performed to verify the former’s performance. We expect the proposed ETA prediction methodology to predict ETA to help build an intelligent port system.

   @article{park2021vessel,
     keywords = {Vessel ETA prediction, AIS data, Reinforcement learning, Markov Chain, Bayesian sampling, Port operations},
     doi = {https://doi.org/10.1016/j.martra.2021.100012},
     title = {Vessel estimated time of arrival prediction system based on a path-finding algorithm},
     author = {Park, Kikun and Sim, Sunghyun and Bae, Hyerim},
     journal = {Maritime Transport Research},
     volume = {2},
     pages = {100012},
     year = {2021},
     publisher = {Elsevier}
   }

   Keywords: Vessel ETA prediction AIS data Reinforcement learning Markov Chain Bayesian sampling Port operations
2. Appl. Sci.

   Parallel-structure deep learning for prediction of remaining time of process instances

   Nur Ahmad Wahid, Hyerim Bae, Taufik Nur Adi, and 2 more authors

   Applied Sciences, 2021

   PDF Bib

   @article{wahid2021parallel,
     title = {Parallel-structure deep learning for prediction of remaining time of process instances},
     author = {Wahid, Nur Ahmad and Bae, Hyerim and Adi, Taufik Nur and Choi, Yulim and Iskandar, Yelita Anggiane},
     keywords = {Remaining time prediction, Deep learning, Process mining, Predictive process monitoring},
     journal = {Applied Sciences},
     volume = {11},
     number = {21},
     pages = {9848},
     year = {2021},
     publisher = {MDPI}
   }

   Keywords: Remaining time prediction Deep learning Process mining Predictive process monitoring
3. Appl. Sci.

   Container volume prediction using time-series decomposition with a long short-term memory models

   Eunju Lee, Dohee Kim, and Hyerim Bae

   Applied Sciences, 2021

   PDF Bib

   @article{lee2021container,
     title = {Container volume prediction using time-series decomposition with a long short-term memory models},
     author = {Lee, Eunju and Kim, Dohee and Bae, Hyerim},
     keywords = {Container volume prediction, Time series decomposition, LSTM, Maritime logistics, Time Series Forecasting},
     journal = {Applied Sciences},
     volume = {11},
     number = {19},
     pages = {8995},
     year = {2021},
     publisher = {MDPI}
   }

   Keywords: Container volume prediction Time series decomposition LSTM Maritime logistics Time Series Forecasting
4. IEEE TSC

   Bagging recurrent event imputation for repair of imperfect event log with missing categorical events

   Sunghyun Sim, Hyerim Bae, and Ling Liu

   IEEE Transactions on Services Computing. JCR Q1, Top 7.5%!!! , 2021

   PDF Bib

   @article{sim2021bagging,
     title = {Bagging recurrent event imputation for repair of imperfect event log with missing categorical events},
     author = {Sim, Sunghyun and Bae, Hyerim and Liu, Ling},
     keywords = {Event log imputation, Bagging, RNN, Process mining, Missing data},
     journal = {IEEE Transactions on Services Computing},
     volume = {16},
     number = {1},
     pages = {108--121},
     year = {2021},
     publisher = {IEEE},
   }

   Keywords: Event log imputation Bagging RNN Process mining Missing data
5. Processes

   Interterminal truck routing optimization using cooperative multiagent deep reinforcement learning

   Taufik Nur Adi, Hyerim Bae, and Yelita Anggiane Iskandar

   Processes, 2021

   PDF Bib

   @article{adi2021interterminal,
     title = {Interterminal truck routing optimization using cooperative multiagent deep reinforcement learning},
     author = {Adi, Taufik Nur and Bae, Hyerim and Iskandar, Yelita Anggiane},
     keywords = {Inter-terminal truck routing, Reinforcement learning, Deep learning, Container terminal operations, Optimization},
     journal = {Processes},
     volume = {9},
     number = {10},
     pages = {1728},
     year = {2021},
     publisher = {MDPI}
   }

   Keywords: Inter-terminal truck routing Reinforcement learning Deep learning Container terminal operations Optimization
6. Processes

   Residual life prediction for induction furnace by sequential encoder with s-convolutional LSTM

   Yulim Choi, Hyeunho Kwun, Dohee Kim, and 2 more authors

   Processes, 2021

   PDF Bib

   @article{choi2021residual,
     title = {Residual life prediction for induction furnace by sequential encoder with s-convolutional LSTM},
     author = {Choi, Yulim and Kwun, Hyeunho and Kim, Dohee and Lee, Eunju and Bae, Hyerim},
     keywords = {Remaining Useful Life Prediction, Induction furnace, Sequential encoder, LSTM, Predictive maintenance},
     journal = {Processes},
     volume = {9},
     number = {7},
     pages = {1121},
     year = {2021},
     publisher = {MDPI}
   }

   Keywords: Remaining Useful Life Prediction Induction furnace Sequential encoder LSTM Predictive maintenance
7. ICIC-EL

   Adversarial multivariate regression and auto-regression for predicting completion time of container terminals operation

   Imam Mustafa Kamal and Hyerim Bae

   ICIC Express Letters, 2021

   PDF Bib

   @article{kamal2021adversarial,
     title = {Adversarial multivariate regression and auto-regression for predicting completion time of container terminals operation},
     author = {Kamal, Imam Mustafa and Bae, Hyerim},
     keywords = {Adversarial learning, Multivariate regression, Container terminal operations, Completion time prediction, Deep learning},
     journal = {ICIC Express Letters},
     volume = {5},
     pages = {737--746},
     year = {2021}
   }

   Keywords: Adversarial learning Multivariate regression Container terminal operations Completion time prediction Deep learning
8. ICIC-B

   Prediction of the quay crane’s handling time with external handling factors

   Eunju Lee, Kikun Park, Dohee Kim, and 2 more authors

   ICIC Exp. Lett., B, Appl., 2021

   PDF Bib

   @article{lee2021prediction,
     title = {Prediction of the quay crane's handling time with external handling factors},
     author = {Lee, Eunju and Park, Kikun and Kim, Dohee and Bae, Hyerim and Hong, Changwoo},
     keywords = {Quay crane, Handling time prediction, External factors, Container terminal operations, Machine learning},
     journal = {ICIC Exp. Lett., B, Appl.},
     volume = {12},
     pages = {351--358},
     year = {2021}
   }

   Keywords: Quay crane Handling time prediction External factors Container terminal operations Machine learning
9. 한국빅데이터학회

   CNN 기반의 준지도학습을 활용한 GPR 이미지 분류

   Hyemee Kim and Hyerim Bae

   The Journal of Bigdata, 2021

   Bib

   @article{김혜미2021cnn,
     title = {CNN 기반의 준지도학습을 활용한 GPR 이미지 분류},
     author = {Kim, Hyemee and Bae, Hyerim},
     journal = {The Journal of Bigdata},
     volume = {6},
     number = {1},
     pages = {197--206},
     year = {2021},
     keywords = {CNN, Semi-supervised learning, GPR image, Image classification, Machine learning}
   }

   Keywords: CNN Semi-supervised learning GPR image Image classification Machine learning
10. ICICIC

    Prediction of vessel arrival time using auto identification system data

    Hyeunho Kwun and Hyerim Bae

    International Journal of Innovative Computing, Information and Control, 2021

    PDF Bib

    @article{kwun2021prediction,
      title = {Prediction of vessel arrival time using auto identification system data},
      author = {Kwun, Hyeunho and Bae, Hyerim},
      keywords = {Vessel arrival time, AIS data, Maritime logistics, Prediction model, Time series},
      journal = {International Journal of Innovative Computing, Information and Control},
      volume = {17},
      number = {3},
      pages = {725--740},
      year = {2021}
    }

    Keywords: Vessel arrival time AIS data Maritime logistics Prediction model Time series
11. Springer

    Incorporating transaction lifecycle information in blockchain process discovery

    Noppon Yaumil Wirawan, Bernardo Nugroho Yahya, and Hyerim Bae

    In Blockchain Technology for IoT Applications, 2021

    PDF Bib

    @inproceedings{wirawan2021blockchain,
      title = {Incorporating transaction lifecycle information in blockchain process discovery},
      author = {Wirawan, Noppon Yaumil and Yahya, Bernardo Nugroho and Bae, Hyerim},
      keywords = {Blockchain, Process discovery, Transaction lifecycle, Smart contracts, Process mining},
      booktitle = {Blockchain Technology for IoT Applications},
      pages = {175--200},
      year = {2021},
      publisher = {Springer}
    }

    Keywords: Blockchain Process discovery Transaction lifecycle Smart contracts Process mining

2020

1. IEEE

   Trajectory linkage and spreader centrality for social epidemic networks

   Imam Mustafa Kamal, Hyerim Bae, and Ki Haing Cho

   IEEE Access, 2020

   PDF Bib

   @article{kamal2020trajectory,
     title = {Trajectory linkage and spreader centrality for social epidemic networks},
     author = {Kamal, Imam Mustafa and Bae, Hyerim and Cho, Ki Haing},
     keywords = {Epidemic networks, Trajectory linkage, Spreader centrality, Social network analysis, Deep learning},
     journal = {IEEE Access},
     volume = {8},
     pages = {210922--210934},
     year = {2020},
     publisher = {IEEE}
   }

   Keywords: Epidemic networks Trajectory linkage Spreader centrality Social network analysis Deep learning
2. Appl. Sci.

   Mining shift work operation from event logs

   Nur Ichsan Utama, Riska Asriana Sutrisnowati, Imam Mustafa Kamal, and 2 more authors

   Applied Sciences, 2020

   PDF Bib

   @article{utama2020mining,
     title = {Mining shift work operation from event logs},
     author = {Utama, Nur Ichsan and Sutrisnowati, Riska Asriana and Kamal, Imam Mustafa and Bae, Hyerim and Park, You-Jin},
     keywords = {Process mining, Shift work operation, Event log analysis, Workflow pattern, Manufacturing},
     journal = {Applied Sciences},
     volume = {10},
     number = {20},
     pages = {7202},
     year = {2020},
     publisher = {MDPI}
   }

   Keywords: Process mining Shift work operation Event log analysis Workflow pattern Manufacturing
3. ICIC-B

   Q-learning-based technique for reduction of number of empty-truck trips in inter-terminal transportation

   Taufik Nur Adi, Yelita Anggiane Iskandar, and Hyerim Bae

   ICIC Express Letters, Part B: Applications, 2020

   PDF Bib

   @article{adi2020q,
     title = {Q-learning-based technique for reduction of number of empty-truck trips in inter-terminal transportation},
     author = {Adi, Taufik Nur and Iskandar, Yelita Anggiane and Bae, Hyerim},
     keywords = {Reinforcement learning, Inter-terminal transportation, Empty truck reduction, Q-learning, Container terminal operations},
     journal = {ICIC Express Letters, Part B: Applications},
     volume = {11},
     number = {10},
     pages = {987--994},
     year = {2020}
   }

   Keywords: Reinforcement learning Inter-terminal transportation Empty truck reduction Q-learning Container terminal operations
4. Appl. Sci.

   DERN: Deep ensemble learning model for short-and long-term prediction of baltic dry index

   Imam Mustafa Kamal, Hyerim Bae, Sunghyun Sim, and 1 more author

   Applied Sciences, 2020

   PDF Bib

   @article{kamal2020dern,
     title = {DERN: Deep ensemble learning model for short-and long-term prediction of baltic dry index},
     author = {Kamal, Imam Mustafa and Bae, Hyerim and Sim, Sunghyun and Yun, Heesung},
     keywords = {Baltic Dry Index, Deep ensemble learning, Time Series Forecasting, Maritime economics, LSTM},
     journal = {Applied Sciences},
     volume = {10},
     number = {4},
     pages = {1504},
     year = {2020},
     publisher = {MDPI}
   }

   Keywords: Baltic Dry Index Deep ensemble learning Time Series Forecasting Maritime economics LSTM
5. Neurocomputing

   Data pixelization for predicting completion time of events

   Imam Mustafa Kamal, Hyerim Bae, Nur Ichsan Utama, and 1 more author

   Neurocomputing. JCR Q1, Top 18.14%!!! , 2020

   PDF Bib

   @article{kamal2020data,
     title = {Data pixelization for predicting completion time of events},
     author = {Kamal, Imam Mustafa and Bae, Hyerim and Utama, Nur Ichsan and Choi, Yulim},
     keywords = {Data pixelization, Completion time prediction, Deep learning, Process mining, Image encoding},
     journal = {Neurocomputing},
     volume = {374},
     pages = {64--76},
     year = {2020},
     publisher = {Elsevier}
   }

   Keywords: Data pixelization Completion time prediction Deep learning Process mining Image encoding
6. 한국빅데이터논문지

   글로벌 팬데믹 상황에서의 긴급지원금 예산 배분 정책에 대한 연구

   Kikun Park, Dohee Kim, Seulgi Kim, and 2 more authors

   한국빅데이터논문지, 2020

   Bib

   @article{박기군2020글로벌,
     title = {글로벌 팬데믹 상황에서의 긴급지원금 예산 배분 정책에 대한 연구},
     author = {Park, Kikun and Kim, Dohee and Kim, Seulgi and Choi, Jiwon and Bae, Hyerim},
     journal = {한국빅데이터논문지},
     year = {2020},
     volume = {5},
     number = {2},
     keywords = {Global pandemic, Emergency relief fund, Budget allocation, Policy analysis, COVID-19}
   }

   Keywords: Global pandemic Emergency relief fund Budget allocation Policy analysis COVID-19
7. 한국빅데이터논문지

   시계열 군집을 활용한 부산시 감염병 지원 정책방향: COVID-19 사례를 중심으로

   Hyeunho Kwun, Dohee Kim, Chan-ho Park, and 3 more authors

   한국빅데이터논문지, 2020

   Bib

   @article{권현호2020시계열,
     title = {시계열 군집을 활용한 부산시 감염병 지원 정책방향: COVID-19 사례를 중심으로},
     author = {Kwun, Hyeunho and Kim, Dohee and Park, Chan-ho and Lee, Eun-joo and Cho, Gi-hyung and Bae, Hyerim},
     journal = {한국빅데이터논문지},
     volume = {5},
     number = {1},
     year = {2020},
     keywords = {Time Series Forecasting, COVID-19, Infectious disease policy, Busan, Public health}
   }

   Keywords: Time Series Forecasting COVID-19 Infectious disease policy Busan Public health
8. BigComp

   Gene expression prediction using stacked temporal convolutional network

   Imam Mustafa Kamal, Nur Ahmad Wahid, and Hyerim Bae

   In 2020 IEEE International Conference on Big Data and Smart Computing (BigComp), 2020

   PDF Bib

   @inproceedings{kamal2020gene,
     title = {Gene expression prediction using stacked temporal convolutional network},
     author = {Kamal, Imam Mustafa and Wahid, Nur Ahmad and Bae, Hyerim},
     keywords = {Gene expression prediction, Temporal convolutional network, Deep learning, Bioinformatics, Time series},
     booktitle = {2020 IEEE International Conference on Big Data and Smart Computing (BigComp)},
     pages = {454--460},
     year = {2020},
     organization = {IEEE}
   }

   Keywords: Gene expression prediction Temporal convolutional network Deep learning Bioinformatics Time series
9. BigComp

   Method of predictive maintenance for induction furnace based on neural network

   Sunghyun Sim, Hyeunho Kwun, Dohee Kim, and 2 more authors

   In 2020 IEEE International Conference on Big Data and Smart Computing (BigComp), 2020

   PDF Bib

   @inproceedings{sim2020predictive,
     title = {Method of predictive maintenance for induction furnace based on neural network},
     author = {Sim, Sunghyun and Kwun, Hyeunho and Kim, Dohee and Lee, Eunju and Bae, Hyerim},
     keywords = {Predictive maintenance, Induction furnace, Neural network, Deep learning, Manufacturing},
     booktitle = {2020 IEEE International Conference on Big Data and Smart Computing (BigComp)},
     pages = {609--612},
     year = {2020},
     organization = {IEEE}
   }

   Keywords: Predictive maintenance Induction furnace Neural network Deep learning Manufacturing
10. Sensors

    Interterminal truck routing optimization using deep reinforcement learning

    Taufik Nur Adi, Yelita Anggiane Iskandar, and Hyerim Bae

    Sensors, 2020

    PDF Bib

    @article{adi2020interterminal,
      title = {Interterminal truck routing optimization using deep reinforcement learning},
      author = {Adi, Taufik Nur and Iskandar, Yelita Anggiane and Bae, Hyerim},
      keywords = {Inter-terminal truck routing, Deep reinforcement learning, Container terminal operations, Optimization, Multi-agent system},
      journal = {Sensors},
      volume = {20},
      number = {20},
      pages = {5794},
      year = {2020},
      publisher = {MDPI}
    }

    Keywords: Inter-terminal truck routing Deep reinforcement learning Container terminal operations Optimization Multi-agent system
11. IEEE BigComp

    Container-Based Coloured Petri-Net Co-simulation Framework

    Iq Reviessay Pulshashi, Hyerim Bae, and Riska Asriana Sutrisnowati

    In 2020 IEEE International Conference on Big Data and Smart Computing (BigComp), 2020

    PDF Bib

    @inproceedings{pulshashi2020container,
      title = {Container-Based Coloured Petri-Net Co-simulation Framework},
      author = {Pulshashi, Iq Reviessay and Bae, Hyerim and Sutrisnowati, Riska Asriana},
      keywords = {Colored Petri-net, Co-simulation, Container-based, Process simulation, Distributed computing},
      booktitle = {2020 IEEE International Conference on Big Data and Smart Computing (BigComp)},
      pages = {601--604},
      year = {2020},
      organization = {IEEE}
    }

    Keywords: Colored Petri-net Co-simulation Container-based Process simulation Distributed computing