Hai Liu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Hai Liu has a total of 65 co-authors. Here are the top co-authors by number of shared publications:
- Tianyong Hao
South China Normal University
13 shared publications - Fu Lee Wang
Hong Kong Metropolitan University, Hong Kong, China
9 shared publications - Kairong Hu
School of Computer Science, South China Normal University, Guangzhou, China
4 shared publications - Chaobo He
South China Normal University
3 shared publications - Guanhua Chen
South China Normal University
3 shared publications - Yong Tang
South China Normal University
3 shared publications
Latest Publications
- 20 July 2026
- Published by Springer Nature
- p. 241-252
PaR: Prompt-as-Expert Routing with Context-Aware Fusion for Efficient Visual Document Understanding
- 20 July 2026
- Published by Springer Nature
- p. 241-252
PaR: Prompt-as-Expert Routing with Context-Aware Fusion for Efficient Visual Document Understanding
- 12 May 2026
- Published by Springer Nature
- p. 289-304
HFTS: Time-Span-Aware Historical–Future Modeling for Temporal Knowledge Graph Completion
- 12 May 2026
- Published by Springer Nature
- p. 289-304
HFTS: Time-Span-Aware Historical–Future Modeling for Temporal Knowledge Graph Completion
- 10 November 2025
- Published by Springer Nature
- p. 100-114
LQRPP-KGQA: LLM-Based Question Reasoning Paths Prediction For Enhancing Question-Specific Subgraph Construction
- 10 November 2025
- Published by Springer Nature
- p. 100-114
LQRPP-KGQA: LLM-Based Question Reasoning Paths Prediction For Enhancing Question-Specific Subgraph Construction
- 7 August 2025
- Published by Springer Nature in Applied Spatial Analysis and Policy
- Vol. 18 (3) , 1-22
Subjective Perception and Spatial Equity: Geographic Characteristics of Educational Divergence in Guangdong Province
- 7 August 2025
- Published by Springer Nature in Applied Spatial Analysis and Policy
- Vol. 18 (3) , 1-22
Subjective Perception and Spatial Equity: Geographic Characteristics of Educational Divergence in Guangdong Province
- 23 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Information Forensics and Security
- Vol. 20 (15566013) , 5465-5477
Diverse Co-Saliency Feature Learning for Text-Based Person Retrieval
Text-based Person Retrieval (TPR) plays a pivotal role in video surveillance systems for safeguarding public safety. As a fine-grained retrieval task, TPR faces the significant challenge of precisely capturing highly discriminative features across image and text modalities. Existing methods primarily focus on establishing modality-shared feature spaces to bridge cross-modal discrepancies. However, these methods are prone to disturbances from irrelevant information, such as background noises in the visual modality, and often over-emphasize specific local regions while neglecting the capture of diverse discriminative modal features, thereby limiting the robustness of cross-modal matching. In this paper, we introduce a novel framework, termed the Diverse Co-saliency Feature Learning Network (DCFL), which mines the co-saliency information between image and text modalities and enhances the diversity of cross-modal discriminative features while mitigating the interference of noise. Specifically, to construct cross-modal co-saliency features, we devise the Intra-modal Saliency Feature Learning (ISFL) and Cross-modal Saliency Feature Matching (CSFM) modules. ISFL employs a weighted mask mechanism to guide the model in reducing the impact of noise information in both modalities. Complementing ISFL, CSFM establishes consistent relationships between saliency features across modalities, leveraging text descriptions to align pedestrian-relevant visual regions. Furthermore, we propose the Diverse Co-saliency Feature Mining (DCFM) to bolster the diversity of discriminative co-saliency features across both image and text modalities. This module integrates a diversity regularization term, enabling the extraction of varied visual cues and capturing comprehensive features of the target individual. Extensive benchmark experiments demonstrate a substantial superiority of our approach over the state-of-the-art methods. Our code is available at https://github.com/ysh-strive/DCFL
- 23 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Information Forensics and Security
- Vol. 20 (15566013) , 5465-5477
Diverse Co-Saliency Feature Learning for Text-Based Person Retrieval
Text-based Person Retrieval (TPR) plays a pivotal role in video surveillance systems for safeguarding public safety. As a fine-grained retrieval task, TPR faces the significant challenge of precisely capturing highly discriminative features across image and text modalities. Existing methods primarily focus on establishing modality-shared feature spaces to bridge cross-modal discrepancies. However, these methods are prone to disturbances from irrelevant information, such as background noises in the visual modality, and often over-emphasize specific local regions while neglecting the capture of diverse discriminative modal features, thereby limiting the robustness of cross-modal matching. In this paper, we introduce a novel framework, termed the Diverse Co-saliency Feature Learning Network (DCFL), which mines the co-saliency information between image and text modalities and enhances the diversity of cross-modal discriminative features while mitigating the interference of noise. Specifically, to construct cross-modal co-saliency features, we devise the Intra-modal Saliency Feature Learning (ISFL) and Cross-modal Saliency Feature Matching (CSFM) modules. ISFL employs a weighted mask mechanism to guide the model in reducing the impact of noise information in both modalities. Complementing ISFL, CSFM establishes consistent relationships between saliency features across modalities, leveraging text descriptions to align pedestrian-relevant visual regions. Furthermore, we propose the Diverse Co-saliency Feature Mining (DCFM) to bolster the diversity of discriminative co-saliency features across both image and text modalities. This module integrates a diversity regularization term, enabling the extraction of varied visual cues and capturing comprehensive features of the target individual. Extensive benchmark experiments demonstrate a substantial superiority of our approach over the state-of-the-art methods. Our code is available at https://github.com/ysh-strive/DCFL
- Open Access
- 8 May 2025
- Published by Association for Computing Machinery (ACM)
- p. 2923-2926
From Literature to Lab: Hardware-Independent Autonomous Chemical Synthesis with Reinforcement Learning
- 8 May 2025
- Published by Association for Computing Machinery (ACM)
- p. 2923-2926
Open AccessFrom Literature to Lab: Hardware-Independent Autonomous Chemical Synthesis with Reinforcement Learning
- 22 September 2024
- Published by Springer Nature
- p. 116-131
Improving Dialogue State Tracking with Interactive Acts Attention and Attention Divergence Loss Function
- 22 September 2024
- Published by Springer Nature
- p. 116-131
Improving Dialogue State Tracking with Interactive Acts Attention and Attention Divergence Loss Function
- 21 August 2024
- Published by Springer Nature in World Wide Web
- Vol. 27 (5) , 1-17
Multi-stage enhanced representation learning for document reranking based on query view
- 21 August 2024
- Published by Springer Nature in World Wide Web
- Vol. 27 (5) , 1-17
Multi-stage enhanced representation learning for document reranking based on query view
- 18 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Emerging Topics in Computational Intelligence
- Vol. 8 (3) , 2640-2653
Improving Topic Tracing with a Textual Reader for Conversational Knowledge Based Question Answering
Conversational KBQA(Knowledge Based Question Answering) is a sequential question-answering process in the form of conversation based on knowledge, and it has been paid great attention in recent years. One of the major challenges in conversational KBQA is the ellipsis and co-reference of topic entities in follow-up questions, which affects the performance of the whole conversational KBQA. Previous approaches identified the topics of current turn questions by encoding conversation records or modeling entities in conversation records. However, they ignored the meanings carried by the entities themselves in the modeling process. To solve the above problem and mitigate the impact of the problem on the whole KBQA system, we propose a new textual reader to integrate entity-related textual information and construct a graph-based neural network containing the textual reader to determine the topics of questions. The graph-based neural network scores entities in each question in conversations. Further, the scores are jointly cooperated with the similarity between questions and answers to obtain the correct answers in conversational KBQA systems. Our proposed method improved the accuracy with 5.5% at topic entity prediction and 1.5% at conversational KBQA on benchmark datasets compared with baseline methods in more real-world settings respectively. Experiment results on two datasets demonstrate that our proposed method improves the performance of topic tracing and conversational KBQA.
- 18 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Emerging Topics in Computational Intelligence
- Vol. 8 (3) , 2640-2653
Improving Topic Tracing with a Textual Reader for Conversational Knowledge Based Question Answering
Conversational KBQA(Knowledge Based Question Answering) is a sequential question-answering process in the form of conversation based on knowledge, and it has been paid great attention in recent years. One of the major challenges in conversational KBQA is the ellipsis and co-reference of topic entities in follow-up questions, which affects the performance of the whole conversational KBQA. Previous approaches identified the topics of current turn questions by encoding conversation records or modeling entities in conversation records. However, they ignored the meanings carried by the entities themselves in the modeling process. To solve the above problem and mitigate the impact of the problem on the whole KBQA system, we propose a new textual reader to integrate entity-related textual information and construct a graph-based neural network containing the textual reader to determine the topics of questions. The graph-based neural network scores entities in each question in conversations. Further, the scores are jointly cooperated with the similarity between questions and answers to obtain the correct answers in conversational KBQA systems. Our proposed method improved the accuracy with 5.5% at topic entity prediction and 1.5% at conversational KBQA on benchmark datasets compared with baseline methods in more real-world settings respectively. Experiment results on two datasets demonstrate that our proposed method improves the performance of topic tracing and conversational KBQA.
- 19 February 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Consumer Electronics
- Vol. 70 (1) , 2806-2816
Improving Open Intent Detection via Triplet-Contrastive Learning and Adaptive Boundary
Open intent detection is a critical task within dialogue systems, aiming to effectively classify known intents while also identifying unknown intents that have not been encountered in the training data. Learning discriminative representations and precise decision boundary are two key challenges in open intent detection. To address these challenges, this paper proposes a Triplet-Contrastive representation learning strategy and an Adaptive Boundary (TCAB) method. Traditional methods often confine the features of known intents to compact regions, assuming that open intents exist outside these regions. Nevertheless, open intents can be dispersed within the known intents. To tackle this issue, this paper introduces a triplet-contrastive representation learning method to acquire discriminative semantic features and distinguish between similar open intents and known intents. Additionally, to achieve more precise decision boundaries, an adaptive boundary method takes into account both in-class and out-of-class instances for calibrating boundary radius. Comprehensive experiments demonstrate that our approach yields substantial improvements over a range of baseline methods on three benchmark datasets. Our code is available at https://github.com/cgh-code777/TCAB.
- 19 February 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Consumer Electronics
- Vol. 70 (1) , 2806-2816
Improving Open Intent Detection via Triplet-Contrastive Learning and Adaptive Boundary
Open intent detection is a critical task within dialogue systems, aiming to effectively classify known intents while also identifying unknown intents that have not been encountered in the training data. Learning discriminative representations and precise decision boundary are two key challenges in open intent detection. To address these challenges, this paper proposes a Triplet-Contrastive representation learning strategy and an Adaptive Boundary (TCAB) method. Traditional methods often confine the features of known intents to compact regions, assuming that open intents exist outside these regions. Nevertheless, open intents can be dispersed within the known intents. To tackle this issue, this paper introduces a triplet-contrastive representation learning method to acquire discriminative semantic features and distinguish between similar open intents and known intents. Additionally, to achieve more precise decision boundaries, an adaptive boundary method takes into account both in-class and out-of-class instances for calibrating boundary radius. Comprehensive experiments demonstrate that our approach yields substantial improvements over a range of baseline methods on three benchmark datasets. Our code is available at https://github.com/cgh-code777/TCAB.