Weidong Zhao
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Weidong Zhao has a total of 8 co-authors. Here are the top co-authors by number of shared publications:
- Gong Gao
Tongji University
2 shared publications - Wenlong Hou
School of Electronic and Information Engineering, Tongji University, Shanghai, China
2 shared publications - Xianhui Liu2 shared publications
- Yiheng Zhuang2 shared publications
- Yueying Liu2 shared publications
- Ning Jia
College of Electronic and Information Engineering, Tongji University, Shanghai, China
1 shared publication
Latest Publications
- 1 December 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Knowledge and Data Engineering
- Vol. 38 (12) , 8279-8292
Improving Generalization and Robustness in Offline Reinforcement Learning via Boundary-Aware Data Augmentation
- 1 December 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Knowledge and Data Engineering
- Vol. 38 (12) , 8279-8292
Improving Generalization and Robustness in Offline Reinforcement Learning via Boundary-Aware Data Augmentation
- 30 July 2026
- Published by Elsevier in Expert Systems with Applications
- Vol. 333
Planning as graphs: Adaptive agentic planning for LLM agents
- 30 July 2026
- Published by Elsevier in Expert Systems with Applications
- Vol. 333
Planning as graphs: Adaptive agentic planning for LLM agents
- Open Access
- 31 March 2026
- Published by Elsevier in SSRN Electronic Journal
Planning as Graphs: Adaptive Agentic Planning for LLM Agents
Planning is fundamental to LLM agents that solve complex, multi-step tasks, yet how to represent plans in a structured form that supports controllable execution remains an open challenge. Existing approaches often express plans as unstructured natural language or linear Chain-of-Thought (CoT), making it difficult to explicitly capture dependency structures among subtasks. They also lack outcome-conditioned transitions such as success/failure branches, limiting structured failure recovery during execution. We propose a Graph-based Adaptive Agentic Planner centered on a unified graph representation, GraphPlan, that jointly encodes prerequisite constraints and outcome-conditioned transitions, providing a structured substrate for failure localization and recovery. From a topological perspective, we show that CoT and unstructured planning are degenerate forms of GraphPlan under different constraints. Moreover, GraphPlan adaptively expands or collapses according to task complexity, enabling on-demand structuring. On SWE-bench Lite, a long-horizon interactive benchmark for repository-level repair, our method achieves a 51% resolution rate, substantially outperforming CoT (38%) and ReAct without planning (28.3%).
- 31 March 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessPlanning as Graphs: Adaptive Agentic Planning for LLM Agents
Planning is fundamental to LLM agents that solve complex, multi-step tasks, yet how to represent plans in a structured form that supports controllable execution remains an open challenge. Existing approaches often express plans as unstructured natural language or linear Chain-of-Thought (CoT), making it difficult to explicitly capture dependency structures among subtasks. They also lack outcome-conditioned transitions such as success/failure branches, limiting structured failure recovery during execution. We propose a Graph-based Adaptive Agentic Planner centered on a unified graph representation, GraphPlan, that jointly encodes prerequisite constraints and outcome-conditioned transitions, providing a structured substrate for failure localization and recovery. From a topological perspective, we show that CoT and unstructured planning are degenerate forms of GraphPlan under different constraints. Moreover, GraphPlan adaptively expands or collapses according to task complexity, enabling on-demand structuring. On SWE-bench Lite, a long-horizon interactive benchmark for repository-level repair, our method achieves a 51% resolution rate, substantially outperforming CoT (38%) and ReAct without planning (28.3%).
- 2 February 2026
- Published by Elsevier in Neural Networks
- Vol. 199, 108667
Improving policy exploitation in online reinforcement learning with instant retrospect action
- 2 February 2026
- Published by Elsevier in Neural Networks
- Vol. 199, 108667
Improving policy exploitation in online reinforcement learning with instant retrospect action