About
I am a Research Scientist at Singapore Management University, working on reinforcement learning for LLM reasoning and agents. I hold a Ph.D. in Computer Science from SMU, an M.S. in Management Science and Engineering from Jilin University, and a B.S. in Information Management and Information Systems from Nanjing Agricultural University.
Research Focus
My work connects sequential decision-making, post-training, and inference-time control. Current directions include:
- LLM reasoning and agents: search-augmented reasoning, agentic AI, and adaptive inference.
- Policy learning: PPO, GRPO, GAIL, and offline RL for reasoning control.
- Optimization: constrained RL, Lagrangian and primal-dual methods, MILP, and Bayesian optimization.
Selected Research
Early Rumor Detection: Learning When a Frozen LLM Should Decide
Co-first-author work accepted at KDD 2026. A lightweight imitation-learning policy learns when to stop reading while balancing accuracy, earliness, and stability.
Selected Publications
- Qian Shao, Pradeep Varakantham, and Shih-Fen Cheng, “Imitating Cost Constrained Behaviors in Reinforcement Learning”, The International Conference on Automated Planning and Scheduling(ICAPS), June 2024, full research paper.
- Qian Shao and Shih-Fen Cheng, “Preference-Aware Delivery Planning for Last-Mile Logistics”, The 22nd International Conference on Autonomous Agents and Multiagent Systems(AAMAS), June 2023, full research paper.
- Qian Shao, Varakantham Pradeep, and Shih-Fen Cheng, “Cost Constrained Imitation Learning”, Adaptive and Learning Agents Workshop at AAMAS(ALA), June 2023, workshop paper.
- Ling Cheng, Qian Shao*, Fengzhu Zeng, and Feida Zhu, “A Full-History Network Dataset for BTC Asset Decentralization Profiling”, 2024 IEEE International Conference on Big Data(IEEE BigData), December 2024, full research paper.
- Ling Cheng, Feida Zhu, Qian Shao, Jiashu Pu, and Fengzhu Zeng, “Early Detection of Malicious Crypto Addresses with Asset Path Tracing and Selection”, IEEE Transactions on Knowledge and Data Engineering (TKDE), 2024.
- Qian Shao, Tien Mai, and Shih-Fen Cheng, “Constrained Pricing in Logit-Based Revenue Management”, INFORMS Journal on Computing, 2026.
- Fengzhu Zeng, Qian Shao, Ling Cheng, Wei Gao, Shih-Fen Cheng, Jing Ma, and Cheng Niu, “LLM-based Few-Shot Early Rumor Detection with Imitation Agent”, KDD 2026.
* denotes equal contribution
News 📢
- [2026] Our work on LLM-based few-shot early rumor detection with imitation learning was accepted at KDD 2026.
[Oct 2025] Joined Singapore Management University as a Research Scientist, working on reinforcement learning for LLM reasoning and agents.
- [Dec 2024] Our Paper “Early Detection of Malicious Crypto Addresses with Asset Path Tracing and Selection” (work with Ling Cheng, Jiashu Pu, Fengzhu Zeng,and Prof. Feida Zhu) is accepted at IEEE Transactions on Knowledge and Data Engineering (TKDE)!
- [Oct 2024] Our Paper “A Full-History Network Dataset for BTC Asset Decentralization Profiling” (work with Ling Cheng, Fengzhu Zeng,and Prof. Feida Zhu) is accepted at IEEE BigData 2024, see you in Washington D.C. in December!
- [Aug 2024] I am awarded the SMU Presidential Doctoral Fellowship in the Academic Year 2024/2025.
- [Jun 2024] Presented our work Imitating Cost Constrained Behaviors in Reinforcement Learning at ICAPS 2024.
