Shuai Pan
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Shuai Pan has a total of 8 co-authors. Here are the top co-authors by number of shared publications:
- Wei Zhang
Advanced Institution of Information Technology Peking University, Hangzhou, China
3 shared publications - Jiayin Chen
Advanced Institution of Information Technology, Peking University, Hangzhou, China
2 shared publications Wei Zhang
Peking University
2 shared publications- Xi Chen
Advanced Institution of Information Technology, Peking University, Hangzhou, China
2 shared publications Yun Fang
Peking University
2 shared publicationsYun Jiang
Peking University
2 shared publications
Latest Publications
- 5 August 2026
- Published by Springer Nature
- p. 233-248
PRISM: Position-Guided Region-Based Image Separation Into Multi-layers
- 5 August 2026
- Published by Springer Nature
- p. 233-248
PRISM: Position-Guided Region-Based Image Separation Into Multi-layers
- 22 June 2025
- Published by Springer Nature
- p. 102-114
E2E-AFG: An End-to-End Model with Adaptive Filtering for Retrieval-Augmented Generation
- 22 June 2025
- Published by Springer Nature
- p. 102-114
E2E-AFG: An End-to-End Model with Adaptive Filtering for Retrieval-Augmented Generation
- Open Access
- 23 April 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 1-12
Hierarchical contrastive learning for multi-label text classification
Multi-label text classification presents a significant challenge within the field of text classification, particularly due to the hierarchical nature of labels, where labels are organized in a tree-like structure that captures parent-child and sibling relationships. This hierarchy reflects semantic dependencies among labels, with higher-level labels representing broader categories and lower-level labels capturing more specific distinctions. Traditional methods often fail to deeply understand and leverage this hierarchical structure, overlooking the subtle semantic differences and correlations that distinguish one label from another. To address this shortcoming, we introduce a novel method called Hierarchical Contrastive Learning for Multi-label Text Classification (HCL-MTC). Our approach leverages the contrastive knowledge embedded within label relationships by constructing a graph representation that explicitly models the hierarchical dependencies among labels. Specifically, we recast multi-label text classification as a multi-task learning problem, incorporating a hierarchical contrastive loss that is computed through a carefully designed sampling process. This unique loss function enables our model to effectively capture both the correlations and distinctions among labels, thereby enhancing the model’s ability to learn the intricacies of the label hierarchy. Experimental results on widely-used datasets, such as RCV1-v2 and WoS, demonstrate that our proposed HCL-MTC model achieves substantial performance gains compared to baseline methods.
- 23 April 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 1-12
Open AccessHierarchical contrastive learning for multi-label text classification
Multi-label text classification presents a significant challenge within the field of text classification, particularly due to the hierarchical nature of labels, where labels are organized in a tree-like structure that captures parent-child and sibling relationships. This hierarchy reflects semantic dependencies among labels, with higher-level labels representing broader categories and lower-level labels capturing more specific distinctions. Traditional methods often fail to deeply understand and leverage this hierarchical structure, overlooking the subtle semantic differences and correlations that distinguish one label from another. To address this shortcoming, we introduce a novel method called Hierarchical Contrastive Learning for Multi-label Text Classification (HCL-MTC). Our approach leverages the contrastive knowledge embedded within label relationships by constructing a graph representation that explicitly models the hierarchical dependencies among labels. Specifically, we recast multi-label text classification as a multi-task learning problem, incorporating a hierarchical contrastive loss that is computed through a carefully designed sampling process. This unique loss function enables our model to effectively capture both the correlations and distinctions among labels, thereby enhancing the model’s ability to learn the intricacies of the label hierarchy. Experimental results on widely-used datasets, such as RCV1-v2 and WoS, demonstrate that our proposed HCL-MTC model achieves substantial performance gains compared to baseline methods.
- Open Access
- 22 July 2024
- Published by Springer Nature in International Journal of Computational Intelligence Systems
- Vol. 17 (1) , 1-13
CLESR: Context-Based Label Knowledge Enhanced Span Recognition for Named Entity Recognition
Named entity recognition (NER) stands as a pivotal task in natural language processing, bearing profound implications for diverse downstream applications. Recent scholarship underscores the substantial performance gains attainable by integrating label knowledge into token representations, effectively treating named entity recognition as a machine reading comprehension task. Nevertheless, extant methodologies often inadequately leverage the potential of label knowledge. In response to this limitation, we introduce the Context-based Label Knowledge Enhanced Span Recognition (CLESR) architecture, designed to augment label knowledge through the assimilation of contextual information. We formulate an annotation paradigm tailored to nested scenarios, concurrently training external context-based label knowledge using conventional word association learning algorithms. The ensuing context-based label knowledge is seamlessly integrated into the model via a dedicated label attention module, thereby fortifying label learning capabilities during training. To adeptly manage both flat and nested entities, we implement a global pointer as our decoding strategy, enabling direct predictions of the positions and corresponding categories of named entities. Rigorous experimentation across six widely recognized benchmarks substantiates the efficacy of CLESR in both flat and nested tasks. Impressively, our model achieves performance enhancements surpassing those of the baseline BERT-MRC model, with gains of + 0.29%, + 0.77%, + 0.39%, + 1.9%, and + 0.47% on English CoNLL2003, English OntoNotes 4.0, Chinese MSRA, English ACE2004, and English ACE2005, respectively.
- 22 July 2024
- Published by Springer Nature in International Journal of Computational Intelligence Systems
- Vol. 17 (1) , 1-13
Open AccessCLESR: Context-Based Label Knowledge Enhanced Span Recognition for Named Entity Recognition
Named entity recognition (NER) stands as a pivotal task in natural language processing, bearing profound implications for diverse downstream applications. Recent scholarship underscores the substantial performance gains attainable by integrating label knowledge into token representations, effectively treating named entity recognition as a machine reading comprehension task. Nevertheless, extant methodologies often inadequately leverage the potential of label knowledge. In response to this limitation, we introduce the Context-based Label Knowledge Enhanced Span Recognition (CLESR) architecture, designed to augment label knowledge through the assimilation of contextual information. We formulate an annotation paradigm tailored to nested scenarios, concurrently training external context-based label knowledge using conventional word association learning algorithms. The ensuing context-based label knowledge is seamlessly integrated into the model via a dedicated label attention module, thereby fortifying label learning capabilities during training. To adeptly manage both flat and nested entities, we implement a global pointer as our decoding strategy, enabling direct predictions of the positions and corresponding categories of named entities. Rigorous experimentation across six widely recognized benchmarks substantiates the efficacy of CLESR in both flat and nested tasks. Impressively, our model achieves performance enhancements surpassing those of the baseline BERT-MRC model, with gains of + 0.29%, + 0.77%, + 0.39%, + 1.9%, and + 0.47% on English CoNLL2003, English OntoNotes 4.0, Chinese MSRA, English ACE2004, and English ACE2005, respectively.
- Open Access
- 21 August 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 11, 90554-90562
Solving Data Imbalance in Text Classification With Constructing Contrastive Samples
Contrastive learning (CL) has been successfully applied in Natural Language Processing (NLP) as a powerful representation learning method and has shown promising results in various downstream tasks. Recent research has highlighted the importance of constructing effective contrastive samples through data augmentation. However, current data augmentation methods primarily rely on random word deletion, substitution, and cropping, which may introduce noisy samples and hinder representation learning. In this article, we propose a novel approach to address data imbalance in text classification by constructing contrastive samples. Our method involves the use of a Label-indicative Component to generate high-quality positive samples for the minority class, along with the introduction of a Hard Negative Mixing strategy to synthesize challenging negative samples at the feature level. By applying supervised contrastive learning to these samples, we are able to obtain superior text representations, which significantly benefit text classification tasks with imbalanced data. Our approach effectively mitigates distributional biases and promotes noise-resistant representation learning. To validate the effectiveness of our method, we conducted experiments on benchmark datasets (THUCNews, AG’s News, 20NG) as well as the imbalanced FDCNews dataset. The code for our method is publicly available at the following GitHub repository: https://github.com/hanggun/CLDMTC.
- 21 August 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 11, 90554-90562
Open AccessSolving Data Imbalance in Text Classification With Constructing Contrastive Samples
Contrastive learning (CL) has been successfully applied in Natural Language Processing (NLP) as a powerful representation learning method and has shown promising results in various downstream tasks. Recent research has highlighted the importance of constructing effective contrastive samples through data augmentation. However, current data augmentation methods primarily rely on random word deletion, substitution, and cropping, which may introduce noisy samples and hinder representation learning. In this article, we propose a novel approach to address data imbalance in text classification by constructing contrastive samples. Our method involves the use of a Label-indicative Component to generate high-quality positive samples for the minority class, along with the introduction of a Hard Negative Mixing strategy to synthesize challenging negative samples at the feature level. By applying supervised contrastive learning to these samples, we are able to obtain superior text representations, which significantly benefit text classification tasks with imbalanced data. Our approach effectively mitigates distributional biases and promotes noise-resistant representation learning. To validate the effectiveness of our method, we conducted experiments on benchmark datasets (THUCNews, AG’s News, 20NG) as well as the imbalanced FDCNews dataset. The code for our method is publicly available at the following GitHub repository: https://github.com/hanggun/CLDMTC.