Lihui Cen
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Lihui Cen has a total of 72 co-authors. Here are the top co-authors by number of shared publications:
- Xiaofang Chen
School of Automation, Central South University, Changsha, China
14 shared publications - Yongfang Xie
Central South University
11 shared publications - Jinping Liu
Hunan Normal University
6 shared publications - Shichao Kan
School of Computer Science and Engineering, Central South University, Changsha, Hunan, China
6 shared publications - Weihua Gui
Central South University
4 shared publications - Yigang Cen
Beijing Jiaotong University
4 shared publications
Latest Publications
- 29 September 2026
- Published by Elsevier in Image and Vision Computing
- Vol. 176
IVCP: Instruction-Guided Video Context Prompting for video action recognition
- 29 September 2026
- Published by Elsevier in Image and Vision Computing
- Vol. 176
IVCP: Instruction-Guided Video Context Prompting for video action recognition
- Open Access
- 14 April 2026
- Published by Elsevier in SSRN Electronic Journal
IVCP: Instruction-Guided Video Context Prompting for Video Action Recognition
Video recognition is a fundamental task in high-level video understanding. However, bridging semantic intent and visual perception remains challenging. Traditional label-supervised methods rely on annotations and therefore lack rich spatio-temporal reasoning. Multimodal Large Language Models (MLLMs) offer instruction-following flexibility. However, they often struggle to extract effective visual representations. Recent prompt-based approaches improve adaptability but still depend on static, task-agnostic prompts initialized from templates, GPT-like texts, or visual prototypes, which limits dynamic temporal reasoning. To address these limitations, we propose a spatio-temporal decoding framework via Instruction-Vision-Conditioned Prompts (IVCP), which leverages a large language model (LLM) to interpret task instructions and generate instruction-aware prompts conditioned on textual and visual contexts. Integrated into a transformer-based decoder, IVCP enables instruction-grounded spatio-temporal feature extraction and achieves superior accuracy across NTU-60/120, UCF-101, HMDB-51 and Something Something V2(SthSthV2) benchmarks.
- 14 April 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessIVCP: Instruction-Guided Video Context Prompting for Video Action Recognition
Video recognition is a fundamental task in high-level video understanding. However, bridging semantic intent and visual perception remains challenging. Traditional label-supervised methods rely on annotations and therefore lack rich spatio-temporal reasoning. Multimodal Large Language Models (MLLMs) offer instruction-following flexibility. However, they often struggle to extract effective visual representations. Recent prompt-based approaches improve adaptability but still depend on static, task-agnostic prompts initialized from templates, GPT-like texts, or visual prototypes, which limits dynamic temporal reasoning. To address these limitations, we propose a spatio-temporal decoding framework via Instruction-Vision-Conditioned Prompts (IVCP), which leverages a large language model (LLM) to interpret task instructions and generate instruction-aware prompts conditioned on textual and visual contexts. Integrated into a transformer-based decoder, IVCP enables instruction-grounded spatio-temporal feature extraction and achieves superior accuracy across NTU-60/120, UCF-101, HMDB-51 and Something Something V2(SthSthV2) benchmarks.
- 15 March 2026
- Published by Elsevier in Expert Systems with Applications
- Vol. 319
Progressive anchoring-driven consistent dual reconstruction for cross-domain open-mode process monitoring
- 15 March 2026
- Published by Elsevier in Expert Systems with Applications
- Vol. 319
Progressive anchoring-driven consistent dual reconstruction for cross-domain open-mode process monitoring
- 15 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Fuzzy Systems
- Vol. 34 (2) , 580-592
Efficiently Data-Driven Offline Generation and Online Modification of Belief Rules
Belief rule based classification system (BRBCS) is a useful model to handle classification problems. However, there are two problems that limit its applications. On the one hand, existing non-optimization based offline generation method can not fully extract information contained in historical data. On the other hand, there is lack of non-optimization based online modification method. In some applications, optimization based online modification can be inapplicable because of its high computational expense. Non-optimization based online modification is indispensable. In this paper, the efficient data-driven methods of belief rule offline generation and online modification are proposed, which belong to non-optimization based scope. The former changes traditional one-by-one observation into batch-by-batch observation. It directly generates a belief rule from a batch of historical data without batch fusing, so that more information can be extracted from a batch of historical data. The latter is based on multi-weight extension and actual meaning of rule parameter. The multi-weight extension is to handle the drawback that single weight is easy to cause quality degeneration of belief rules during process of online modification. The actual meaning of rule parameter is used to establish a heuristic mechanism of online parameter adjustment. The related experiments have demonstrated the effectiveness and advancement of proposed methods. BRBCS equipped with proposed methods can become more useful in some industrial applications.
- 15 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Fuzzy Systems
- Vol. 34 (2) , 580-592
Efficiently Data-Driven Offline Generation and Online Modification of Belief Rules
Belief rule based classification system (BRBCS) is a useful model to handle classification problems. However, there are two problems that limit its applications. On the one hand, existing non-optimization based offline generation method can not fully extract information contained in historical data. On the other hand, there is lack of non-optimization based online modification method. In some applications, optimization based online modification can be inapplicable because of its high computational expense. Non-optimization based online modification is indispensable. In this paper, the efficient data-driven methods of belief rule offline generation and online modification are proposed, which belong to non-optimization based scope. The former changes traditional one-by-one observation into batch-by-batch observation. It directly generates a belief rule from a batch of historical data without batch fusing, so that more information can be extracted from a batch of historical data. The latter is based on multi-weight extension and actual meaning of rule parameter. The multi-weight extension is to handle the drawback that single weight is easy to cause quality degeneration of belief rules during process of online modification. The actual meaning of rule parameter is used to establish a heuristic mechanism of online parameter adjustment. The related experiments have demonstrated the effectiveness and advancement of proposed methods. BRBCS equipped with proposed methods can become more useful in some industrial applications.
- Open Access
- 27 October 2025
- Published by Association for Computing Machinery (ACM)
- p. 7500-7509
Contrastive Regularization over LoRA for Multimodal Biomedical Image Incremental Learning
- 27 October 2025
- Published by Association for Computing Machinery (ACM)
- p. 7500-7509
Open AccessContrastive Regularization over LoRA for Multimodal Biomedical Image Incremental Learning
- 15 September 2025
- Published by Springer Nature in JOM
- Vol. 77 (11) , 9094-9112
MSTFNet: A Multi-scale Time-Frequency Domain Network for Current Efficiency Prediction in Aluminum Electrolysis Process
- 15 September 2025
- Published by Springer Nature in JOM
- Vol. 77 (11) , 9094-9112
MSTFNet: A Multi-scale Time-Frequency Domain Network for Current Efficiency Prediction in Aluminum Electrolysis Process
- 3 September 2025
- Published by Elsevier in Engineering Applications of Artificial Intelligence
- Vol. 161
Unknown fault incremental learning based on shapelet prototypical network for streaming industrial signals
- 3 September 2025
- Published by Elsevier in Engineering Applications of Artificial Intelligence
- Vol. 161
Unknown fault incremental learning based on shapelet prototypical network for streaming industrial signals
- 13 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Informatics
- Vol. 21 (11) , 8562-8573
Zero-Shot Fault Diagnosis in Industrial Processes Using Graph-Regularized Coupled Dictionary Learning
In real-world industrial processes, it is crucial to avoid damaging faults to ensure steady operations. However, new faulty conditions can arise in variable production environments. Traditional fault diagnosis methods perform poorly with few or no fault data for training. To address this issue, we propose a graph-regularized coupled dictionary learning (GRCDL) method for zero-shot fault diagnosis in industrial processes. Our method enhances the coupled dictionary learning (CDL) framework by integrating class prototype learning, which utilizes the rich information of seen class faults in the data feature space. A feature correlation function is also designed to facilitate the transformation and reconstruction of data and semantic features in the shared space, preventing certain classes from becoming the nearest neighbors of unrelated classes in a low-dimensional semantic space. To mitigate noise interference and preserve local manifold structures, the graph-regularized constraint is developed to enhance the robustness and locality of the coupled dictionaries. Considering the different underlying distributions between unseen and seen faults, a novel domain adaptation function is proposed. This function uses the learned coupled dictionaries as mediums for knowledge transfer and update from seen to unseen classes and effectively learns the class prototypes of unseen faults. Extensive experiments in industrial processes verify the superior performance of our method for zero-shot fault diagnosis.
- 13 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Informatics
- Vol. 21 (11) , 8562-8573
Zero-Shot Fault Diagnosis in Industrial Processes Using Graph-Regularized Coupled Dictionary Learning
In real-world industrial processes, it is crucial to avoid damaging faults to ensure steady operations. However, new faulty conditions can arise in variable production environments. Traditional fault diagnosis methods perform poorly with few or no fault data for training. To address this issue, we propose a graph-regularized coupled dictionary learning (GRCDL) method for zero-shot fault diagnosis in industrial processes. Our method enhances the coupled dictionary learning (CDL) framework by integrating class prototype learning, which utilizes the rich information of seen class faults in the data feature space. A feature correlation function is also designed to facilitate the transformation and reconstruction of data and semantic features in the shared space, preventing certain classes from becoming the nearest neighbors of unrelated classes in a low-dimensional semantic space. To mitigate noise interference and preserve local manifold structures, the graph-regularized constraint is developed to enhance the robustness and locality of the coupled dictionaries. Considering the different underlying distributions between unseen and seen faults, a novel domain adaptation function is proposed. This function uses the learned coupled dictionaries as mediums for knowledge transfer and update from seen to unseen classes and effectively learns the class prototypes of unseen faults. Extensive experiments in industrial processes verify the superior performance of our method for zero-shot fault diagnosis.
- 1 July 2025
- Published by Springer Nature in Signal, Image and Video Processing
- Vol. 19 (10) , 1-12
Enhanced-MOT: Robust Multi-Pedestrian Tracking with Enhanced IoU and Deep Features Association
- 1 July 2025
- Published by Springer Nature in Signal, Image and Video Processing
- Vol. 19 (10) , 1-12
Enhanced-MOT: Robust Multi-Pedestrian Tracking with Enhanced IoU and Deep Features Association
- 14 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-14
Small Sample-Oriented Variable Working Condition Fault Diagnosis via Nondata-Enhanced Multicategory Contrastive Learning
In real-world industrial scenarios, production equipment and process systems usually operate under variable working conditions, and abnormal/faulty samples are hard to collect, bringing great challenges in implementing intelligent fault diagnosis. This work presents a novel data augmentation-free multi-category contrastive learning framework specifically designed for small sample-oriented fault diagnosis under variable working conditions. The contributions are twofold. First, a sample tuple batch construction strategy (STBCS) is proposed to establish sample tuple batches without resorting to data augmentation. This is achieved by leveraging positive sample pairs from all fault categories within the limited small-sample training set, thereby enhancing the diversity and representativeness of the training data. Second, an adaptive multi-category contrastive learning method (AMCCLM) is proposed based on the constructed sample tuples to learn high-quality domain-invariant features, thereby enhancing the generalizability and robustness of the fault diagnosis system. Extensive confirmatory and comparative experiments were carried out on two publically available benchmark datasets: the Spectra Quest (SQ) and the Case Western Reserve University (CWRU) bearing dataset, and a real industrial application scenario: the axial flow fan fault diagnosis from a steel-making plant. The experimental results demonstrate the superiority and robustness of the proposed method compared to representative state-of-the-art methods, especially under conditions involving small samples and variable working conditions.
- 14 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-14
Small Sample-Oriented Variable Working Condition Fault Diagnosis via Nondata-Enhanced Multicategory Contrastive Learning
In real-world industrial scenarios, production equipment and process systems usually operate under variable working conditions, and abnormal/faulty samples are hard to collect, bringing great challenges in implementing intelligent fault diagnosis. This work presents a novel data augmentation-free multi-category contrastive learning framework specifically designed for small sample-oriented fault diagnosis under variable working conditions. The contributions are twofold. First, a sample tuple batch construction strategy (STBCS) is proposed to establish sample tuple batches without resorting to data augmentation. This is achieved by leveraging positive sample pairs from all fault categories within the limited small-sample training set, thereby enhancing the diversity and representativeness of the training data. Second, an adaptive multi-category contrastive learning method (AMCCLM) is proposed based on the constructed sample tuples to learn high-quality domain-invariant features, thereby enhancing the generalizability and robustness of the fault diagnosis system. Extensive confirmatory and comparative experiments were carried out on two publically available benchmark datasets: the Spectra Quest (SQ) and the Case Western Reserve University (CWRU) bearing dataset, and a real industrial application scenario: the axial flow fan fault diagnosis from a steel-making plant. The experimental results demonstrate the superiority and robustness of the proposed method compared to representative state-of-the-art methods, especially under conditions involving small samples and variable working conditions.