Publications

Peer-reviewed papers, accepted conference papers, and public arXiv preprints.

* Equal contribution (co-first authors)

Beyond Corrected Memory: Execution Consistency in Multi-Agent Systems

Zhe Yu*, Zixuan Wang*, Peidong Wang, Hehai Lin, Ruochen Zhao, Chengwei Qin

arXiv preprint arXiv:2610.08101 2026 · Submitted to ICLR 2027

Defines execution consistency beyond accurate shared memory and studies evidence-based diagnosis and recovery in multi-agent workflows.

Paper

MAS-Gate: Unlocking Multi-Agent System Lifecycle via Node-Centric Quantification

Hehai Lin*, Zixuan Wang*, Ming Yan*, Peidong Wang, Shilei Cao, Minzhi Li, Bosheng Ding, Juepeng Zheng, Chengwei Qin

arXiv preprint arXiv:2610.17613 2026 · Submitted to ICLR 2027

Uses node-level utility and interaction strength to guide multi-agent design, runtime editing, and post-hoc error attribution.

Paper

Degrees of Freedom in Multi-Agent Coordination: Between Centralized and Decentralized Regimes

Yiqing Peng, Hehai Lin, Zixuan Wang, Dongyi Zhang, Zhengjie Zhu, Yi Han, Yangyang Zheng, Hao Wang, Qian Li, Chengwei Qin

arXiv preprint arXiv:2610.17751 2026 · Submitted to ICLR 2027

Studies a continuous coordination space between centralized and decentralized multi-agent systems through delegated decision freedom.

Paper

AMA: Adaptive Memory via Multi-Agent Collaboration

Weiquan Huang*, Zixuan Wang*, Hehai Lin, Sudong Wang, Bo Xu, Qian Li, Beier Zhu, Linyi Yang, Chengwei Qin

Findings of the Association for Computational Linguistics: ACL 2026

Coordinates agents to construct, retrieve, verify, and refresh long-term memory across multiple granularities.

BizCompass: Benchmarking the Reasoning Capabilities of LLMs in Business Knowledge and Applications

Jianing Hao, Yuhe Wu, Yuanjian Xu, Shichang Meng, Shuai Yuan, Wei Zeng, Zixuan Wang, Guang Zhang

Findings of the Association for Computational Linguistics: ACL 2026

Benchmarks how foundational business knowledge supports LLM reasoning in practical applications.

PaperDOI

T-Mem: Memory That Anticipates, Not Archives

Weidong Guo, Dakai Wang, Zixuan Wang, Hui Liu, Yu Xu

EMNLP 2026 Main Conference (accepted)

Long-term conversational memory that prepares associative as well as descriptive retrieval cues at write time.

PaperCode

Unified-MAS: Universally Generating Domain-Specific Nodes for Empowering Automatic Multi-Agent Systems

Hehai Lin, Yu Yan, Zixuan Wang, Bo Xu, Sudong Wang, Weiquan Huang, Ruochen Zhao, Minzhi Li, Chengwei Qin

EMNLP 2026 Main Conference (accepted)

Generates and optimizes domain-specific nodes to improve automatic multi-agent system construction.

PaperCode

UNGIA: Unnoticeable Injection Attack on Graph Neural Networks

Zixuan Wang, N. Chen, L. Y. Zhang, D. Chen

International Joint Conference on Neural Networks (IJCNN 2026)

Studies stealthy graph injection attacks using representative-node selection and bi-level optimization to balance attack effectiveness with low detectability.