Yabin Wang
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
Yabin Wang has a total of 44 co-authors. Here are the top co-authors by number of shared publications:
- Xiaopeng Hong
Harbin Institute of Technology, Harbin, China
20 shared publications - Zhiheng Ma
Faculty of Computility Microelectronics, Shenzhen University of Advanced Technology, Shenzhen, China
10 shared publications - Zhou Su
Xi’an Jiaotong University
6 shared publications - Yaowei Wang
Harbin Institute of Technology (Shenzhen), Shenzhen, China and Pengcheng Laboratory, Shenzhen, China
4 shared publications - Zhiwu Huang
University of Southampton United Kingdom
4 shared publications - Zhengqin Lai
Harbin Institute of Technology, Shenzhen, China and Pengcheng Laboratory, Shenzhen, China
3 shared publications
Latest Publications
- Open Access
- 9 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 17673-17680
Multi-Granularity Deepfake Detector-grounded Progressive Explanation Generation for Explainable Deepfake Detection
- 9 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 17673-17680
Open AccessMulti-Granularity Deepfake Detector-grounded Progressive Explanation Generation for Explainable Deepfake Detection
- Open Access
- 9 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 17517-17524
EgoAffect-R1: Elimination-Guided Reinforcement Learning for Egocentric Embodied Emotional Reasoning
Egocentric social-affective understanding is central to embodied affective intelligence, yet remains challenging for multimodal large language models because correct interpretations often depend on subtle egocentric evidence and fine-grained distinctions among plausible candidates. We propose EgoAffect-R1, a process-aware and elimination-guided reinforcement learning framework for egocentric embodied emotional reasoning. We construct EgoAffect-CoT-18k for cold-start CoT supervision and EgoAffect-RL-5k for verifiable reinforcement learning. Our EgoAffect reward scores format, question type, emotional tone, and elimination-based candidate discrimination, penalizing trajectories that discard the correct semantic option. To reduce option-order shortcuts, we introduce Option-Perturbed DAPO, which trains on reordered candidates and aligns rewards with their semantic identities. Experiments on EgoLink show that EgoAffect-R1 achieves 92.75% accuracy on the EgoLink validation split and 92.46% accuracy in the EgoLink 2026 Challenge. These results show that structured cold-start data, elimination-guided rewards, and option-perturbed RL improve egocentric affective reasoning. Code and data are available at https://github.com/ZhengQinLai/EgoAffect-R1.
- 9 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 17517-17524
Open AccessEgoAffect-R1: Elimination-Guided Reinforcement Learning for Egocentric Embodied Emotional Reasoning
Egocentric social-affective understanding is central to embodied affective intelligence, yet remains challenging for multimodal large language models because correct interpretations often depend on subtle egocentric evidence and fine-grained distinctions among plausible candidates. We propose EgoAffect-R1, a process-aware and elimination-guided reinforcement learning framework for egocentric embodied emotional reasoning. We construct EgoAffect-CoT-18k for cold-start CoT supervision and EgoAffect-RL-5k for verifiable reinforcement learning. Our EgoAffect reward scores format, question type, emotional tone, and elimination-based candidate discrimination, penalizing trajectories that discard the correct semantic option. To reduce option-order shortcuts, we introduce Option-Perturbed DAPO, which trains on reordered candidates and aligns rewards with their semantic identities. Experiments on EgoLink show that EgoAffect-R1 achieves 92.75% accuracy on the EgoLink validation split and 92.46% accuracy in the EgoLink 2026 Challenge. These results show that structured cold-start data, elimination-guided rewards, and option-perturbed RL improve egocentric affective reasoning. Code and data are available at https://github.com/ZhengQinLai/EgoAffect-R1.
- 1 July 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Pattern Analysis and Machine Intelligence
- Vol. 48 (7) , 7707-7719
Penny-Wise and Pound-Foolish in AI-Generated Image Detection
The rise of AI-generated images has sparked serious concerns about their potential misuse across various domains, prompting the urgent need for robust detection methods. Despite advancements, many current approaches prioritize short-term gains at the expense of long-term effectiveness. This paper critiques the overly specialized approach of fine-tuning pre-trained models for short-term gains on a single AI image dataset, while disregarding the long-term imperative of achieving generalization and knowledge retention. To address this trade-off issue, we propose a novel learning framework (PoundNet) for the generalization of AI-generated image detection on a pre-trained vision-language model. PoundNet incorporates a learnable prompt design and a balanced objective to preserve broad knowledge from upstream tasks (object classification) while enhancing generalization for downstream tasks (AI-generated image detection). We train PoundNet on a single standard AI image dataset, following common practice in the literature. We then evaluate its performance across 10 large-scale public AI-generated image detection datasets with 5 main evaluation metrics, forming the largest benchmark test set for assessing the generalization ability of AI-generated image detection models, to our knowledge. The comprehensive benchmark evaluation demonstrates that PoundNet successfully balances generalization with knowledge retention, achieving a remarkable relative improvement of 19% in AI-generated image detection performance compared to state-of-the-art methods, while maintaining a strong performance of 63% on object classification tasks.
- 1 July 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Pattern Analysis and Machine Intelligence
- Vol. 48 (7) , 7707-7719
Penny-Wise and Pound-Foolish in AI-Generated Image Detection
The rise of AI-generated images has sparked serious concerns about their potential misuse across various domains, prompting the urgent need for robust detection methods. Despite advancements, many current approaches prioritize short-term gains at the expense of long-term effectiveness. This paper critiques the overly specialized approach of fine-tuning pre-trained models for short-term gains on a single AI image dataset, while disregarding the long-term imperative of achieving generalization and knowledge retention. To address this trade-off issue, we propose a novel learning framework (PoundNet) for the generalization of AI-generated image detection on a pre-trained vision-language model. PoundNet incorporates a learnable prompt design and a balanced objective to preserve broad knowledge from upstream tasks (object classification) while enhancing generalization for downstream tasks (AI-generated image detection). We train PoundNet on a single standard AI image dataset, following common practice in the literature. We then evaluate its performance across 10 large-scale public AI-generated image detection datasets with 5 main evaluation metrics, forming the largest benchmark test set for assessing the generalization ability of AI-generated image detection models, to our knowledge. The comprehensive benchmark evaluation demonstrates that PoundNet successfully balances generalization with knowledge retention, achieving a remarkable relative improvement of 19% in AI-generated image detection performance compared to state-of-the-art methods, while maintaining a strong performance of 63% on object classification tasks.
- 3 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Affective Computing
- Vol. PP (3) , 1-14
Incremental Micro-Expression Recognition: A Benchmark
Micro-expression recognition (MER) plays a pivotal role in understanding hidden emotions. While traditional methods assume static datasets, real-world scenarios require adapting to continuously evolving data streams. To this end, we introduce the first benchmark specifically designed for Incremental Micro Expression Recognition (IMER). Our contributions include: Firstly, we formulate a composite class-domain incremental learning setting and construct a sequential benchmark from five representative datasets with carefully curated learning orders to reflect real-world scenarios. Secondly, we establish robust evaluation protocols with a fold-binding strategy to ensure rigorous and feasible cross-session validation, using comprehensive metrics and novel cross-domain visualizations to diagnose performance. Thirdly, we propose Mahalanobis Refinement (MR), a two-stage approach that leverages accumulated second-order statistics for stability and Mahalanobis-constrained refinement for plasticity. Extensive experiments demonstrate that MR significantly outper forms state-of-the-art baselines, effectively balancing the stability plasticity dilemma. This work lays the foundation for scalable and adaptive micro-expression analysis. All source codes will be released.
- 3 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Affective Computing
- Vol. PP (3) , 1-14
Incremental Micro-Expression Recognition: A Benchmark
Micro-expression recognition (MER) plays a pivotal role in understanding hidden emotions. While traditional methods assume static datasets, real-world scenarios require adapting to continuously evolving data streams. To this end, we introduce the first benchmark specifically designed for Incremental Micro Expression Recognition (IMER). Our contributions include: Firstly, we formulate a composite class-domain incremental learning setting and construct a sequential benchmark from five representative datasets with carefully curated learning orders to reflect real-world scenarios. Secondly, we establish robust evaluation protocols with a fold-binding strategy to ensure rigorous and feasible cross-session validation, using comprehensive metrics and novel cross-domain visualizations to diagnose performance. Thirdly, we propose Mahalanobis Refinement (MR), a two-stage approach that leverages accumulated second-order statistics for stability and Mahalanobis-constrained refinement for plasticity. Extensive experiments demonstrate that MR significantly outper forms state-of-the-art baselines, effectively balancing the stability plasticity dilemma. This work lays the foundation for scalable and adaptive micro-expression analysis. All source codes will be released.
- 9 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Multimedia
- Vol. PP (15209210) , 1-13
Continual Conceptual Entity Learning for Text-to-Image Generative Models
Current Text-to-Image generative models struggle to continuously learn multiple distinct entities or concepts, limiting their scalability and hindering practical deployment in dynamic environments. We formulate this task as Continual Conceptual Entity Learning (CEL) and propose a novel framework called Continual Entity Adapter Learning (CEAL). CEAL leverages a compact set of tunable parameters, termed SuperLoRA, to efficient and scalable learning of new entities. We propose a dynamic rank-increasing strategy to train the SuperLoRA, balancing computational efficiency with performance. To evaluate our method, we create three benchmarks encompassing generic objects, human faces, and artistic styles. Experimental results demonstrate that CEAL effectively learns new entities while preserving prior knowledge, outperforming existing methods in both entity fidelity and parameter efficiency.
- 9 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Multimedia
- Vol. PP (15209210) , 1-13
Continual Conceptual Entity Learning for Text-to-Image Generative Models
Current Text-to-Image generative models struggle to continuously learn multiple distinct entities or concepts, limiting their scalability and hindering practical deployment in dynamic environments. We formulate this task as Continual Conceptual Entity Learning (CEL) and propose a novel framework called Continual Entity Adapter Learning (CEAL). CEAL leverages a compact set of tunable parameters, termed SuperLoRA, to efficient and scalable learning of new entities. We propose a dynamic rank-increasing strategy to train the SuperLoRA, balancing computational efficiency with performance. To evaluate our method, we create three benchmarks encompassing generic objects, human faces, and artistic styles. Experimental results demonstrate that CEAL effectively learns new entities while preserving prior knowledge, outperforming existing methods in both entity fidelity and parameter efficiency.
- Top Cited Papers
- 4 March 2026
- Published by Elsevier in Pattern Recognition
- Vol. 178
Linguistic profiling of deepfakes: An open database for next-Generation deepfake detection
- 4 March 2026
- Published by Elsevier in Pattern Recognition
- Vol. 178
Top Cited PapersLinguistic profiling of deepfakes: An open database for next-Generation deepfake detection
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Multimedia
- Vol. 28, 13518-13529
Spectral Interaction Modeling for Hyperspectral Image Classification
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Multimedia
- Vol. 28, 13518-13529
Spectral Interaction Modeling for Hyperspectral Image Classification
- 20 November 2025
- Published by Elsevier in Pattern Recognition
- Vol. 172
Asymmetric modal fusion for multi-modal crowd counting
- 20 November 2025
- Published by Elsevier in Pattern Recognition
- Vol. 172
Asymmetric modal fusion for multi-modal crowd counting
- 20 November 2025
- Published by Elsevier in Pattern Recognition
- Vol. 172
Dual-Attention based prompt generation and catalyzing for instance-wise continual learning
- 20 November 2025
- Published by Elsevier in Pattern Recognition
- Vol. 172
Dual-Attention based prompt generation and catalyzing for instance-wise continual learning
- 3 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Artificial Intelligence
- Vol. 6 (5) , 1373-1385
Prompt Customization for Continual Learning
Contemporary continual learning approaches typically select prompts from a pool, which function as supplementary inputs to a pre-trained model. However, this strategy is hindered by the inherent noise of its selection approach when handling increasing tasks. In response to these challenges, we reformulate the prompting approach for continual learning and propose the prompt customization (PC) method. PC mainly comprises a prompt generation module (PGM) and a prompt modulation module (PMM). In contrast to conventional methods that employ hard prompt selection, PGM assigns different coefficients to prompts from a fixed-sized pool of prompts and generates tailored prompts. Moreover, PMM further modulates the prompts by adaptively assigning weights according to the correlations between input data and corresponding prompts. We evaluate our method on four benchmark datasets for three diverse settings, including the class, domain, and task-agnostic incremental learning tasks. Experimental results demonstrate consistent improvement (by up to 16.2%), yielded by the proposed method, over the state-of-the-art (SOTA) techniques. The code has been released online.
- 3 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Artificial Intelligence
- Vol. 6 (5) , 1373-1385
Prompt Customization for Continual Learning
Contemporary continual learning approaches typically select prompts from a pool, which function as supplementary inputs to a pre-trained model. However, this strategy is hindered by the inherent noise of its selection approach when handling increasing tasks. In response to these challenges, we reformulate the prompting approach for continual learning and propose the prompt customization (PC) method. PC mainly comprises a prompt generation module (PGM) and a prompt modulation module (PMM). In contrast to conventional methods that employ hard prompt selection, PGM assigns different coefficients to prompts from a fixed-sized pool of prompts and generates tailored prompts. Moreover, PMM further modulates the prompts by adaptively assigning weights according to the correlations between input data and corresponding prompts. We evaluate our method on four benchmark datasets for three diverse settings, including the class, domain, and task-agnostic incremental learning tasks. Experimental results demonstrate consistent improvement (by up to 16.2%), yielded by the proposed method, over the state-of-the-art (SOTA) techniques. The code has been released online.