Epistemic Exploration for Generalizable Planning and Learning in Non-Stationary Stochastic Settings
Rushang Karia · Pulkit Verma · Gaurav Vipat · Siddharth Srivastava
Keywords:
Model-based Reinforcement Learning
Relational Reinforcement Learning
Non-stationary model-learning
Stochastic planning
2023 Contributed Talk
in
Workshop: Generalization in Planning (GenPlan '23)
in
Workshop: Generalization in Planning (GenPlan '23)
Abstract
Reinforcement Learning (RL) provides a convenient framework for sequential decision making when closed-form transition dynamics are unavailable and can frequently change. However, the high sample complexity of RL approaches limits their utility in the real-world. This paper presents an approach that performs meta-level exploration in the space of models and uses the learned models to compute policies. Our approach interleaves learning and planning allowing data-efficient, task-focused sample collection in the presence of non-stationarity. We conduct an empirical evaluation on benchmark domains and show that our approach significantly outperforms baselines in sample complexity and easily adapts to changing transition systems across tasks.
Video
Chat is not available.
Successful Page Load