Xiuying Chen
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Xiuying Chen has a total of 157 co-authors. Here are the top co-authors by number of shared publications:
- Xiangliang Zhang
University of Notre Dame
11 shared publications Zirui Song
Artificial Intelligence (MBZUAI Zayed University ) ♣ Peking University
9 shared publications- Xin Gao
King Abdullah University of Science and Technology, Thuwal, Saudi Arabia
7 shared publications Zixiang Xu
Artificial Intelligence (MBZUAI Zayed University ) ♣ Peking University
7 shared publications- Mingzhe Li
Ant Group, Beijing, China
6 shared publications - Rui Yan
School of Artificial Intelligence, Wuhan University, Wuhan, China
6 shared publications
Latest Publications
- Open Access
- 9 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 4308-4317
M3MAD-Bench: Multi-Dimensional Evaluation of Multi-Agent Debate Across Domains and Modalities
- 9 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 4308-4317
Open AccessM3MAD-Bench: Multi-Dimensional Evaluation of Multi-Agent Debate Across Domains and Modalities
- Open Access
- 21 October 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Information Systems
- Vol. 44  (8) , 1-27
Unifying Search and Recommendation with Dual-View Representation Learning in a Generative Paradigm
Recommender systems and personalized product search are critical components of modern online platforms, where recommender systems proactively surface relevant items while product search allows users to express their needs through queries. Unifying both tasks in a shared model is promising since it can enhance user modeling and item understanding. Previous approaches mainly follow a discriminative paradigm, utilizing shared encoders to process input features and task-specific heads to perform each task. However, this paradigm faces two key challenges: gradient conflict and manual architecture design overhead. From the information theory perspective, these challenges potentially both stem from the same issue—low mutual information between input features and task-specific output during optimization. To address this, we propose GenSR, a novel generative paradigm for unifying search and recommendation (S&R), which leverages task-specific prompts to induce task-conditioned optimization subspaces in the shared model, thereby enhancing mutual information. To construct effective subspaces for each task, GenSR first prepares informative representations for each subspace and then optimizes both subspaces in one unified model. Specifically, GenSR (1) Dual Representation Learning, which independently models collaborative and semantic historical information to derive expressive item representations and (2) S&R Task Unifying, which utilizes contrastive learning together with instruction tuning to generate task-specific outputs effectively. Extensive experiments on three benchmarks show GenSR outperforms state-of-the-art methods across S&R tasks.
- 21 October 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Information Systems
- Vol. 44  (8) , 1-27
Open AccessUnifying Search and Recommendation with Dual-View Representation Learning in a Generative Paradigm
Recommender systems and personalized product search are critical components of modern online platforms, where recommender systems proactively surface relevant items while product search allows users to express their needs through queries. Unifying both tasks in a shared model is promising since it can enhance user modeling and item understanding. Previous approaches mainly follow a discriminative paradigm, utilizing shared encoders to process input features and task-specific heads to perform each task. However, this paradigm faces two key challenges: gradient conflict and manual architecture design overhead. From the information theory perspective, these challenges potentially both stem from the same issue—low mutual information between input features and task-specific output during optimization. To address this, we propose GenSR, a novel generative paradigm for unifying search and recommendation (S&R), which leverages task-specific prompts to induce task-conditioned optimization subspaces in the shared model, thereby enhancing mutual information. To construct effective subspaces for each task, GenSR first prepares informative representations for each subspace and then optimizes both subspaces in one unified model. Specifically, GenSR (1) Dual Representation Learning, which independently models collaborative and semantic historical information to derive expressive item representations and (2) S&R Task Unifying, which utilizes contrastive learning together with instruction tuning to generate task-specific outputs effectively. Extensive experiments on three benchmarks show GenSR outperforms state-of-the-art methods across S&R tasks.
- 25 August 2026
- Published by Springer Nature in International Journal of Machine Learning and Cybernetics
- Vol. 17  (9) , 437
AI security beyond core domains: resume screening as a case study of adversarial vulnerabilities in specialized LLM applications
- 25 August 2026
- Published by Springer Nature in International Journal of Machine Learning and Cybernetics
- Vol. 17  (9) , 437
AI security beyond core domains: resume screening as a case study of adversarial vulnerabilities in specialized LLM applications
- Open Access
- 8 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 7600-7611
One Model, Multiple Goals: Adaptive Multi-Objective Learning for E-commerce Dialogue Systems
- 8 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 7600-7611
Open AccessOne Model, Multiple Goals: Adaptive Multi-Objective Learning for E-commerce Dialogue Systems
- Open Access
- 8 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 8636-8647
AutoDavis: Automatic and Dynamic Evaluation Protocol of Large Vision-Language Models on Visual Question-Answering
- 8 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 8636-8647
Open AccessAutoDavis: Automatic and Dynamic Evaluation Protocol of Large Vision-Language Models on Visual Question-Answering
- Open Access
- 4 August 2026
- Published by Springer Nature in npj Digital Medicine
MedKGent: a large language model agent framework for constructing temporally evolving medical knowledge graph
- 4 August 2026
- Published by Springer Nature in npj Digital Medicine
Open AccessMedKGent: a large language model agent framework for constructing temporally evolving medical knowledge graph
- Open Access
- 5 June 2026
- Published by MDPI AG in Preprints.org
The Evaluation Bottleneck of Vision-Language-Action Models: A Evaluation-Centric Survey
Vision-Language-Action (VLA) models are advancing faster than the field can evaluate them reliably. Researchers use different metrics and lab-specific protocols, making it hard to tell whether reported gains reflect genuine progress or favorable evaluation choices. We present the first comprehensive benchmark-centric survey of VLA evaluation, covering 582 papers from 2023 to May 2026. We argue that a benchmark number supports a progress claim only if four conditions hold: the benchmark discriminates among top models; metrics capture the claimed capability; the procedure permits cross-paper comparison; and the inference from benchmark to deployment is valid. Current practice fails at all four. Benchmark choice is concentrated and saturated, with leading models clustering near the ceiling of dominant simulation suites. Metric reporting is one-dimensional, dominated by task success rate while efficiency, safety, and trajectory consistency remain underreported. Real-world evaluation is fragmented, with no widely adopted standard and few trials per task. Simulation scores are widely treated as evidence of real-world capability, yet standard task-centric suites correlate only weakly with real-robot performance unless explicitly calibrated to the target physical setup. These failures reveal a fundamental evaluation bottleneck: VLA models are advancing faster than our ability to measure that advance.
- 5 June 2026
- Published by MDPI AG in Preprints.org
Open AccessThe Evaluation Bottleneck of Vision-Language-Action Models: A Evaluation-Centric Survey
Vision-Language-Action (VLA) models are advancing faster than the field can evaluate them reliably. Researchers use different metrics and lab-specific protocols, making it hard to tell whether reported gains reflect genuine progress or favorable evaluation choices. We present the first comprehensive benchmark-centric survey of VLA evaluation, covering 582 papers from 2023 to May 2026. We argue that a benchmark number supports a progress claim only if four conditions hold: the benchmark discriminates among top models; metrics capture the claimed capability; the procedure permits cross-paper comparison; and the inference from benchmark to deployment is valid. Current practice fails at all four. Benchmark choice is concentrated and saturated, with leading models clustering near the ceiling of dominant simulation suites. Metric reporting is one-dimensional, dominated by task success rate while efficiency, safety, and trajectory consistency remain underreported. Real-world evaluation is fragmented, with no widely adopted standard and few trials per task. Simulation scores are widely treated as evidence of real-world capability, yet standard task-centric suites correlate only weakly with real-robot performance unless explicitly calibrated to the target physical setup. These failures reveal a fundamental evaluation bottleneck: VLA models are advancing faster than our ability to measure that advance.
- Open Access
GRASP LoRA: GRPO Guided Adapter Sparsity Policy for Cross Lingual Transfer
Parameter efficient fine tuning is a way to adapt LLMs to new languages when compute or data are limited, yet adapter pipelines usually choose a global prune ratio by grid search. This practice is computationally expensive and development set intensive, since it repeats training, freezes sparsity, and misses fractional optima. We introduce GRASP LoRA (GRPO Guided Adapter Sparsity Policy), which treats global sparsity as a learnable control variable. A GRPO controller interleaves with training, periodically probing candidate prune ratios on a small micro development set and updating a single global prune ratio online from its reward signal. It operates on merged source and target LoRA adapters on a frozen backbone and replaces grid search with one controller run that learns a prune ratio, followed by a single final merge and prune fine tuning run with pruning fixed to that ratio. On cross lingual transfer from English into Arabic and Chinese, including XL-Sum summarization and MLQA extractive question answering with Llama 3 8B, GRASP LoRA improves semantic faithfulness, content coverage, and answer quality over strong target only and merge and prune baselines. It reduces end to end runtime by multiple times relative to grid search, lowers reliance on large development sets, and makes adapter reuse practical for low resource deployment.
Open AccessGRASP LoRA: GRPO Guided Adapter Sparsity Policy for Cross Lingual Transfer
Parameter efficient fine tuning is a way to adapt LLMs to new languages when compute or data are limited, yet adapter pipelines usually choose a global prune ratio by grid search. This practice is computationally expensive and development set intensive, since it repeats training, freezes sparsity, and misses fractional optima. We introduce GRASP LoRA (GRPO Guided Adapter Sparsity Policy), which treats global sparsity as a learnable control variable. A GRPO controller interleaves with training, periodically probing candidate prune ratios on a small micro development set and updating a single global prune ratio online from its reward signal. It operates on merged source and target LoRA adapters on a frozen backbone and replaces grid search with one controller run that learns a prune ratio, followed by a single final merge and prune fine tuning run with pruning fixed to that ratio. On cross lingual transfer from English into Arabic and Chinese, including XL-Sum summarization and MLQA extractive question answering with Llama 3 8B, GRASP LoRA improves semantic faithfulness, content coverage, and answer quality over strong target only and merge and prune baselines. It reduces end to end runtime by multiple times relative to grid search, lowers reliance on large development sets, and makes adapter reuse practical for low resource deployment.
- Open Access
- 27 October 2025
- Published by Association for Computing Machinery (ACM)
- p. 5903-5912
From Individuals to Crowds: Dual-Level Public Response Prediction in Social Media
- 27 October 2025
- Published by Association for Computing Machinery (ACM)
- p. 5903-5912
Open AccessFrom Individuals to Crowds: Dual-Level Public Response Prediction in Social Media
- Open Access
Do LLMs "Feel"? Emotion Circuits Discovery and Control
As the demand for emotional intelligence in large language models (LLMs) grows, a key challenge lies in understanding the internal mechanisms that give rise to emotional expression and in controlling emotions in generated text. This study addresses three core questions: (1) Do LLMs contain context-agnostic mechanisms shaping emotional expression? (2) What form do these mechanisms take? (3) Can they be harnessed for universal emotion control? We first construct a controlled dataset, SEV (Scenario-Event with Valence), to elicit comparable internal states across emotions. Subsequently, we extract context-agnostic emotion directions that reveal consistent, cross-context encoding of emotion (Q1). We identify neurons and attention heads that locally implement emotional computation through analytical decomposition and causal analysis, and validate their causal roles via ablation and enhancement interventions. Next, we quantify each sublayer's causal influence on the model's final emotion representation and integrate the identified local components into coherent global emotion circuits that drive emotional expression (Q2). Directly modulating these circuits achieves 99.65% emotion-expression accuracy on the test set, surpassing prompting- and steering-based methods (Q3). To our knowledge, this is the first systematic study to uncover and validate emotion circuits in LLMs, offering new insights into interpretability and controllable emotional intelligence.
Open AccessDo LLMs "Feel"? Emotion Circuits Discovery and Control
As the demand for emotional intelligence in large language models (LLMs) grows, a key challenge lies in understanding the internal mechanisms that give rise to emotional expression and in controlling emotions in generated text. This study addresses three core questions: (1) Do LLMs contain context-agnostic mechanisms shaping emotional expression? (2) What form do these mechanisms take? (3) Can they be harnessed for universal emotion control? We first construct a controlled dataset, SEV (Scenario-Event with Valence), to elicit comparable internal states across emotions. Subsequently, we extract context-agnostic emotion directions that reveal consistent, cross-context encoding of emotion (Q1). We identify neurons and attention heads that locally implement emotional computation through analytical decomposition and causal analysis, and validate their causal roles via ablation and enhancement interventions. Next, we quantify each sublayer's causal influence on the model's final emotion representation and integrate the identified local components into coherent global emotion circuits that drive emotional expression (Q2). Directly modulating these circuits achieves 99.65% emotion-expression accuracy on the test set, surpassing prompting- and steering-based methods (Q3). To our knowledge, this is the first systematic study to uncover and validate emotion circuits in LLMs, offering new insights into interpretability and controllable emotional intelligence.