Yizhe Yang
I received my Ph.D. in Computer Science from Beijing Institute of Technology in 2026, advised by Prof. Heyan Huang. From 2023 to 2024 I was a visiting Ph.D. student at Singapore Management University, funded by the China Scholarship Council and hosted by Prof. Ee-Peng Lim and Prof. Jing Jiang.
My research centers on conversational agents and LLM agents, in particular how a language model can sustain a long, goal-directed conversation that stays grounded in knowledge, consistent in persona, and strategically sound. Recent work covers counselor agents for motivational interviewing, faithful client and user simulation, and the preference bias that LLM-based dialogue planners silently acquire from simulated rewards.
My work spans both academia and industry: I have published at top-tier AI venues including ACL, EMNLP, AAAI, and TOIS, with representative works such as CAMI, Consistent Client Simulation, Speaker Verification in Agent-Generated Conversations, and MindLLM, the first bilingual foundation model pre-trained end-to-end at a Chinese university. I regularly serve as a reviewer for top-tier conferences and journals, including ICLR, NeurIPS, ICML, ACL ARR, AAAI, CIKM, and Artificial Intelligence.
News
Education
Selected Publications
-
2026MIThinker: A Plug-and-Play Policy-Optimized Thinker for Motivational Interviewing CounselingYizhe Yang, Heyan Huang, Palakorn Achananuparp, Jing Jiang, Ee-Peng LimFindings of ACL 2026CCF-A
-
2026PUPPET: Neural-Symbolic Standardized Patients for Mental HealthCanyu Xu, Yuhang Ji, Zhiyuan Lv, Yi Yi, Yizhe Yang*, Yuanxing Ji, Chong Chen, et al.ACL 2026CCF-A
-
2026Planner Policies are Bias Learners: A Comprehensive Analysis of Preference Bias in LLM-based Dialogue Planner PoliciesYizhe Yang, Heyan Huang, Hang Sun, Junyi Li, Yang Gao
-
2026Simulated Rewards, Skewed Strategies: Tracing the Acquired Preference Bias in LLM-Based Dialogue PlannersHeyan Huang, Yizhe Yang*, Hang Sun, Junyi Li, Yang GaoAAAI 2026CCF-AORAL
-
2026基于联律规则约束解码的对联生成方法杨毅哲, 黄河燕CCL 2026
-
2026Retrieval is NOT Always Needed: Exploring the Timing of Retrieval in Dynamic Retrieval-augmented GenerationYuan Liu, Heyan Huang, Yizhe Yang, Zhen Zeng, Yue Zhou, Zhaoming Wu, Yang GaoExpert Systems with ApplicationsSCI Q1 TOP
-
2025CAMI: A Counselor Agent Supporting Motivational Interviewing through State Inference and Topic ExplorationYizhe Yang, Heyan Huang, Palakorn Achananuparp, Jing Jiang, et al., Ee-Peng LimACL 2025CCF-A
-
2025Consistent Client Simulation for Motivational Interviewing-based CounselingYizhe Yang, Heyan Huang, Palakorn Achananuparp, Jing Jiang, et al., Ee-Peng LimACL 2025CCF-A
-
2025Fundamental Capabilities and Applications of Large Language Models: A SurveyJiawei Li, Yang Gao, Yizhe Yang, et al.ACM Computing SurveysSCI Q1 TOP
-
2025EvoWiki: Evaluating LLMs on Evolving KnowledgeWei Tang, Yixin Cao, Yang Deng, Jiahao Ying, Bo Wang, Yizhe Yang, et al.ACL 2025CCF-A
-
2024Speaker Verification in Agent-Generated ConversationsYizhe Yang, Heyan Huang, Palakorn Achananuparp, Jing Jiang, Ee-Peng LimACL 2024CCF-A
-
2024Building Knowledge-grounded Dialogue Systems with Graph-based Semantic ModellingYizhe Yang, Heyan Huang, Yang Gao, Junyi LiKnowledge-Based SystemsSCI Q1 TOP
-
2024MindLLM: Lightweight Large Language Model Pre-training, Evaluation and Domain ApplicationYizhe Yang, Hang Sun, Junyi Li, Ruikun Liu, Yulan Li, Yuan Liu, Yang Gao, Heyan HuangAI Openmodels
-
2024Automating Dataset Updates Towards Reliable and Timely Evaluation of Large Language ModelsJiahao Ying, Yixin Cao, Yushi Bai, Qianru Sun, Bo Wang, Wei Tang, Zhaojun Ding, Yizhe Yang, et al.NeurIPS 2024CCF-A
-
2024Fundamental Capabilities of Large Language Models and Their Applications in Domain Scenarios: A SurveyJiawei Li, Yizhe Yang, Yu Bai, et al.ACL 2024CCF-A
-
2024Latent Representation Discretization for Unsupervised Text Style GenerationYang Gao, Qianhui Liu, Yizhe Yang, Ke WangInformation Processing & ManagementSCI Q1 TOPESI HIGHLY CITED
-
2023Graph vs. Sequence: An Empirical Study on Knowledge Forms for Knowledge-Grounded DialogueYizhe Yang, Heyan Huang, Yuan Liu, Yang GaoEMNLP 2023
Projects
Core contributor to the first foundation model pre-trained end-to-end at a Chinese university. I worked across data curation, tokenizer design, pre-training, instruction tuning, and domain adaptation. The open-source release has been downloaded more than 30,000 times and has been deployed in pilot applications across government, education, and healthcare.