Xiaoshuai Hao
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Xiaoshuai Hao has a total of 251 co-authors. Here are the top co-authors by number of shared publications:
- Wei Zhou
Cardiff University
6 shared publications Lingfeng Zhang
Tsinghua University
5 shared publicationsLong Chen
Xiaomi EV
5 shared publications- Rong Yin
Beihang University
5 shared publications Can Ma
4 shared publications- Guangyin Jin
Changàn University, Xiàn, China
4 shared publications
Latest Publications
- Open Access
- 9 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 16918-16925
DSBench: A Comprehensive Benchmark for Evaluating External and In-Cabin Risks
- 9 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 16918-16925
Open AccessDSBench: A Comprehensive Benchmark for Evaluating External and In-Cabin Risks
- 1 October 2026
- Published by Elsevier in Information Processing & Management
- Vol. 64 (2)
CAST-DRO: Cognition-aware service triage with configurable risk adjustment for resource-constrained AI information services
- 1 October 2026
- Published by Elsevier in Information Processing & Management
- Vol. 64 (2)
CAST-DRO: Cognition-aware service triage with configurable risk adjustment for resource-constrained AI information services
- Open Access
- 17 September 2026
- Published by Elsevier in SSRN Electronic Journal
Lightweight Causal Target Speaker Extraction via Staged Condition Injection and Learnable Frequency Compression
Causal target speaker extraction (TSE) aims to extract the speech of a target speaker from mixture speech in real time using reference speech and is important in real-time scenarios with overlapping multi-talker speech. Existing causal TSE methods typically encode the reference speech into a global speaker representation, making it difficult to preserve the target speaker's fine-grained structural information along the frequency dimension. Moreover, modeling the full frequency grid incurs high computational costs, while simple frequency compression may weaken local spectral structures and inter-frame dynamic information. To address these issues, this paper proposes a lightweight causal target speaker extraction network, named SCI-LFC, which incorporates three key designs: 1) A speaker encoder is used to generate a target condition that preserves frequency structure, and a staged injection strategy is adopted to introduce this condition only into the last three sequential modules, enabling the model to first learn general acoustic representations and then perform target-related modulation; 2) A learnable local frequency compression strategy is employed, which compresses the frequency dimension through learnable weighted aggregation of adjacent frequency bins, maintaining the frequency-scale correspondence between the target condition and the mixture speech features while lowering the computational overhead caused by full frequency-grid modeling; 3) A lightweight Simplified Temporal Gating (STG) module is introduced for temporal-dynamics compensation. It extracts inter-frame dynamic information from the full-band response and dynamically compensates the intermediate separation features under the compressed frequency representation, thereby improving target speaker extraction performance under causal conditions. Experimental results on the WSJ0-2mix and Aishell-2mix datasets demonstrate that the proposed model achieves a favorable balance between extraction performance and computational efficiency compared with other representative TSE models. Compared with the unprocessed mixture speech, our method achieves SI-SNRi improvements of 14.77 dB and 10.75 dB, PESQ improvements of 1.66 and 1.44, and STOI improvements of 22.42 and 14.68 percentage points, respectively. The inference code and demo examples are available at https://github.com/yuxu93528-ops/SCI-LFC.
- 17 September 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessLightweight Causal Target Speaker Extraction via Staged Condition Injection and Learnable Frequency Compression
Causal target speaker extraction (TSE) aims to extract the speech of a target speaker from mixture speech in real time using reference speech and is important in real-time scenarios with overlapping multi-talker speech. Existing causal TSE methods typically encode the reference speech into a global speaker representation, making it difficult to preserve the target speaker's fine-grained structural information along the frequency dimension. Moreover, modeling the full frequency grid incurs high computational costs, while simple frequency compression may weaken local spectral structures and inter-frame dynamic information. To address these issues, this paper proposes a lightweight causal target speaker extraction network, named SCI-LFC, which incorporates three key designs: 1) A speaker encoder is used to generate a target condition that preserves frequency structure, and a staged injection strategy is adopted to introduce this condition only into the last three sequential modules, enabling the model to first learn general acoustic representations and then perform target-related modulation; 2) A learnable local frequency compression strategy is employed, which compresses the frequency dimension through learnable weighted aggregation of adjacent frequency bins, maintaining the frequency-scale correspondence between the target condition and the mixture speech features while lowering the computational overhead caused by full frequency-grid modeling; 3) A lightweight Simplified Temporal Gating (STG) module is introduced for temporal-dynamics compensation. It extracts inter-frame dynamic information from the full-band response and dynamically compensates the intermediate separation features under the compressed frequency representation, thereby improving target speaker extraction performance under causal conditions. Experimental results on the WSJ0-2mix and Aishell-2mix datasets demonstrate that the proposed model achieves a favorable balance between extraction performance and computational efficiency compared with other representative TSE models. Compared with the unprocessed mixture speech, our method achieves SI-SNRi improvements of 14.77 dB and 10.75 dB, PESQ improvements of 1.66 and 1.44, and STOI improvements of 22.42 and 14.68 percentage points, respectively. The inference code and demo examples are available at https://github.com/yuxu93528-ops/SCI-LFC.
- 8 September 2026
- Published by Elsevier in Information Fusion
- Vol. 139
BiSQAFusion: Multi-level fusion for personalized binaural speech quality assessment in hearing aids
- 8 September 2026
- Published by Elsevier in Information Fusion
- Vol. 139
BiSQAFusion: Multi-level fusion for personalized binaural speech quality assessment in hearing aids
- Open Access
- 2 September 2026
- Published by Springer Nature in Research Square
Beyond Binary Success: A Diagnostic Meta-Evaluation Framework for Fine-Grained Manipulation
Fine-grained manipulation marks a regime where global scene context no longer suffices, and success hinges on the tight coupling of local attribute grounding, high-fidelity spatial perception, and constraint-respecting motor execution. However, current embodied AI benchmarks collapse these capacities into binary success rates, systematically inflating reported capabilities by up to 70% and masking the architectural bottlenecks that impede real-world deployment. We introduce MetaFine, a diagnostic meta-evaluation framework that disentangles manipulation competency along three axes: understanding, perception, and controlled behavior. Built on a compositional task graph, MetaFine absorbs heterogeneous external benchmarks and reconstructs them into diagnostic scenarios of varying complexity under a unified protocol. Evaluating state-of-the-art vision-language-action (VLA) models through this lens exposes severe dimension-specific failures invisible to conventional metrics. Through targeted causal intervention, we identify the visual encoder’s ability to preserve local spatial structure as a key bottleneck for fine-grained precision: improving it directly unlocks previously inaccessible manipulation capabilities without modifying downstream policies. MetaFine further supports hybrid real–sim validation, using limited paired real-world rollouts to calibrate scalable simulation-based estimates for more stable physical benchmarking. By shifting evaluation from ranking to diagnosis, MetaFine turns benchmarking into an actionable compass for repairing the layered capacities underlying genuine physical dexterity. The MetaFine framework, benchmarks, and supporting resources will be publicly released at our project page: https://metafine.github.io/.
- 2 September 2026
- Published by Springer Nature in Research Square
Open AccessBeyond Binary Success: A Diagnostic Meta-Evaluation Framework for Fine-Grained Manipulation
Fine-grained manipulation marks a regime where global scene context no longer suffices, and success hinges on the tight coupling of local attribute grounding, high-fidelity spatial perception, and constraint-respecting motor execution. However, current embodied AI benchmarks collapse these capacities into binary success rates, systematically inflating reported capabilities by up to 70% and masking the architectural bottlenecks that impede real-world deployment. We introduce MetaFine, a diagnostic meta-evaluation framework that disentangles manipulation competency along three axes: understanding, perception, and controlled behavior. Built on a compositional task graph, MetaFine absorbs heterogeneous external benchmarks and reconstructs them into diagnostic scenarios of varying complexity under a unified protocol. Evaluating state-of-the-art vision-language-action (VLA) models through this lens exposes severe dimension-specific failures invisible to conventional metrics. Through targeted causal intervention, we identify the visual encoder’s ability to preserve local spatial structure as a key bottleneck for fine-grained precision: improving it directly unlocks previously inaccessible manipulation capabilities without modifying downstream policies. MetaFine further supports hybrid real–sim validation, using limited paired real-world rollouts to calibrate scalable simulation-based estimates for more stable physical benchmarking. By shifting evaluation from ranking to diagnosis, MetaFine turns benchmarking into an actionable compass for repairing the layered capacities underlying genuine physical dexterity. The MetaFine framework, benchmarks, and supporting resources will be publicly released at our project page: https://metafine.github.io/.
- 6 August 2026
- Published by Elsevier in Information Fusion
- Vol. 138
Parameter-efficient fine-tuning for multimodal fusion: A survey of injection-driven integration mechanisms
- 6 August 2026
- Published by Elsevier in Information Fusion
- Vol. 138
Parameter-efficient fine-tuning for multimodal fusion: A survey of injection-driven integration mechanisms
- Open Access
- 19 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 2150-2161
Question-Adaptive Graph Learning for Multi-hop Retrieval Augmented Generation
- 19 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 2150-2161
Open AccessQuestion-Adaptive Graph Learning for Multi-hop Retrieval Augmented Generation
- Open Access
- 15 July 2026
- Published by Springer Nature in Research Square
Reasoning emerges from constrained inference manifolds in large language models
Reasoning in large language models is predominantly evaluated through labeled benchmarks, conflating task performance with the quality of internal inference. Here we study reasoning as an intrinsic dynamical process by examining the evolution of internal representations during inference. We find that inference-time dynamics consistently self-organize into low-dimensional manifolds embedded within high-dimensional representation spaces. we find that such geometric compression, although pervasive, is not sufficient for stable or reliable reasoning. Instead, effective reasoning dynamics emerge within a constrained structural regime characterized by three conditions: adequate representational expressivity, spontaneous manifold compression, and preservation of non-degenerate information volume within the compressed subspace. Models outside this regime exhibit characteristic pathological inference dynamics. Based on these insights, we introduce a unified, label-free diagnostic computed solely from internal dynamics. These findings suggest that reasoning in LLMs is fundamentally governed by geometric and informational constraints, offering a complementary framework to benchmark-centric assessment.
- 15 July 2026
- Published by Springer Nature in Research Square
Open AccessReasoning emerges from constrained inference manifolds in large language models
Reasoning in large language models is predominantly evaluated through labeled benchmarks, conflating task performance with the quality of internal inference. Here we study reasoning as an intrinsic dynamical process by examining the evolution of internal representations during inference. We find that inference-time dynamics consistently self-organize into low-dimensional manifolds embedded within high-dimensional representation spaces. we find that such geometric compression, although pervasive, is not sufficient for stable or reliable reasoning. Instead, effective reasoning dynamics emerge within a constrained structural regime characterized by three conditions: adequate representational expressivity, spontaneous manifold compression, and preservation of non-degenerate information volume within the compressed subspace. Models outside this regime exhibit characteristic pathological inference dynamics. Based on these insights, we introduce a unified, label-free diagnostic computed solely from internal dynamics. These findings suggest that reasoning in LLMs is fundamentally governed by geometric and informational constraints, offering a complementary framework to benchmark-centric assessment.
- Open Access
- 23 June 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Multimedia Computing, Communications, and Applications
- Vol. 22 (6) , 1-23
DADA++: Dual Alignment Domain Adaptation for Unsupervised Video-Text Retrieval
Video-text retrieval aims at returning the most semantically relevant videos given a textual query, which is a thriving topic in both computer vision and natural language processing communities. This article focuses on a more challenging task, i.e., Unsupervised Domain Adaptation Video-text Retrieval (UDAVR), wherein training and testing data come from different distributions. Previous approaches are mostly derived from classification-based domain adaptation methods, which are neither multi-modal nor suitable for retrieval tasks. They merely alleviate the domain shift while overlooking the pairwise misalignment issue in the target domain, i.e., there exist no semantic relationships between target videos and texts. While Foundation Models like CLIP perform well in in-domain video-text retrieval, their effectiveness significantly drops during domain shifts due to this lack of alignment. To tackle this, we propose a novel method named D ual A lignment D omain A daptation ( DADA ++). Specifically, we first introduce cross-modal semantic embedding to generate discriminative source features in a joint embedding space. Besides, we utilize cross-modal domain adaptations to balance the minimization of domain shift in a smooth manner. Furthermore, we empirically identify the pairwise misalignment in the target domain, and thus propose the i ntegrated D ual A lignment C onsistency (iDAC). The proposed iDAC adaptively aligns the video-text pairs, which are more likely to be relevant in the target domain, by verifying their cross-modal semantic proximity reciprocally in both hard and soft manners. This enables positive pairs to increase progressively while potentially aligning noisy pairs throughout the training procedure. We also provide insights into the functionality of DADA ++ through the lens of Foundation Models, explaining its superiority in a theoretical way. Compared with state-of-the-art methods, DADA ++ achieves 9.4% and 8.5% relative improvements on R@1 under the settings of TGIF \(\rightarrow\) MSR-VTT and TGIF \(\rightarrow\) MSVD, respectively, demonstrating its superior performance.- 23 June 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Multimedia Computing, Communications, and Applications
- Vol. 22 (6) , 1-23
Open AccessDADA++: Dual Alignment Domain Adaptation for Unsupervised Video-Text Retrieval
Video-text retrieval aims at returning the most semantically relevant videos given a textual query, which is a thriving topic in both computer vision and natural language processing communities. This article focuses on a more challenging task, i.e., Unsupervised Domain Adaptation Video-text Retrieval (UDAVR), wherein training and testing data come from different distributions. Previous approaches are mostly derived from classification-based domain adaptation methods, which are neither multi-modal nor suitable for retrieval tasks. They merely alleviate the domain shift while overlooking the pairwise misalignment issue in the target domain, i.e., there exist no semantic relationships between target videos and texts. While Foundation Models like CLIP perform well in in-domain video-text retrieval, their effectiveness significantly drops during domain shifts due to this lack of alignment. To tackle this, we propose a novel method named D ual A lignment D omain A daptation ( DADA ++). Specifically, we first introduce cross-modal semantic embedding to generate discriminative source features in a joint embedding space. Besides, we utilize cross-modal domain adaptations to balance the minimization of domain shift in a smooth manner. Furthermore, we empirically identify the pairwise misalignment in the target domain, and thus propose the i ntegrated D ual A lignment C onsistency (iDAC). The proposed iDAC adaptively aligns the video-text pairs, which are more likely to be relevant in the target domain, by verifying their cross-modal semantic proximity reciprocally in both hard and soft manners. This enables positive pairs to increase progressively while potentially aligning noisy pairs throughout the training procedure. We also provide insights into the functionality of DADA ++ through the lens of Foundation Models, explaining its superiority in a theoretical way. Compared with state-of-the-art methods, DADA ++ achieves 9.4% and 8.5% relative improvements on R@1 under the settings of TGIF \(\rightarrow\) MSR-VTT and TGIF \(\rightarrow\) MSVD, respectively, demonstrating its superior performance. - 11 June 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Image Processing
- Vol. 35 (10577149) , 6289-6304
Synergistic Prompting for Complementarity and Consistency in Incomplete Multi-View Clustering
Incomplete multi-view clustering (IMVC) aims to partition unlabeled multi-view data into semantically coherent groups, even when certain views are missing due to sensor failures, data collection constraints, or privacy concerns. Despite advancements in deep IMVC methods, two critical challenges remain unresolved: (i) the lack of explicit mechanisms to model cross-view complementarity and (ii) the absence of principled strategies to ensure global semantic consistency across views. To address these challenges, we propose SP-IMVC, a novel Synergistic Prompting framework that jointly models complementarity and consistency under view incompleteness. Specifically, we introduce two types of learnable prompts: the Cross-View Complementary Prompt (CVCP), which aggregates auxiliary representations from available views to enrich the semantics of the current view and mitigate information loss; and the Latent Anchor Prompt (LAP), which utilizes a global anchor prompt pool to provide adaptive semantic priors that promote globally consistent representations. These prompts are optimized jointly within a unified architecture to achieve synergistic prompting of cross-view complementarity and global semantic consistency. Extensive experiments on six public benchmarks demonstrate that SP-IMVC consistently outperforms 14 state-of-the-art IMVC approaches, particularly in scenarios with high missing-view ratios, validating the effectiveness and robustness of our synergistic prompt-guided clustering framework. The code will be released to facilitate future research.
- 11 June 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Image Processing
- Vol. 35 (10577149) , 6289-6304
Synergistic Prompting for Complementarity and Consistency in Incomplete Multi-View Clustering
Incomplete multi-view clustering (IMVC) aims to partition unlabeled multi-view data into semantically coherent groups, even when certain views are missing due to sensor failures, data collection constraints, or privacy concerns. Despite advancements in deep IMVC methods, two critical challenges remain unresolved: (i) the lack of explicit mechanisms to model cross-view complementarity and (ii) the absence of principled strategies to ensure global semantic consistency across views. To address these challenges, we propose SP-IMVC, a novel Synergistic Prompting framework that jointly models complementarity and consistency under view incompleteness. Specifically, we introduce two types of learnable prompts: the Cross-View Complementary Prompt (CVCP), which aggregates auxiliary representations from available views to enrich the semantics of the current view and mitigate information loss; and the Latent Anchor Prompt (LAP), which utilizes a global anchor prompt pool to provide adaptive semantic priors that promote globally consistent representations. These prompts are optimized jointly within a unified architecture to achieve synergistic prompting of cross-view complementarity and global semantic consistency. Extensive experiments on six public benchmarks demonstrate that SP-IMVC consistently outperforms 14 state-of-the-art IMVC approaches, particularly in scenarios with high missing-view ratios, validating the effectiveness and robustness of our synergistic prompt-guided clustering framework. The code will be released to facilitate future research.