Co-training for Policy Learning

2019/07/03 by Jialin Song, Song, Jialin, Ravi Lanka +5 · 1 citation
Computer Science · Mathematics · #Artificial Intelligence (cs.AI) #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML) #cs.AI #cs.LG #stat.ML

paper · pdf · doi:10.48550/arxiv.1907.04484

UAI 2019, oral presentation

arxiv created 2019/07/03 · arxiv updated 2019/07/11

Abstract

We study the problem of learning sequential decision-making policies in settings with multiple state-action representations. Such settings naturally arise in many domains, such as planning (e.g., multiple integer programming formulations) and various combinatorial optimization problems (e.g., those with both integer programming and graph-based formulations). Inspired by the classical co-training framework for classification, we study the problem of co-training for policy learning. We present sufficient conditions under which learning from two views can improve upon learning from a single view alone. Motivated by these theoretical insights, we present a meta-algorithm for co-training for sequential decision making. Our framework is compatible with both reinforcement learning and imitation learning. We validate the effectiveness of our approach across a wide range of tasks, including discrete/continuous control and combinatorial optimization.

Cited by

• Guided Policy Optimization under Partial Observability

Related

• Planning with a Learned Policy Basis to Optimally Solve Complex Tasks
• Learning Skills to Navigate without a Master: A Sequential Multi-Policy Reinforcement Learning Algorithm
• From Reinforcement Learning to Optimal Control: A unified framework for sequential decisions
• CORL: Reinforcement Learning of MILP Policies Solved via Branch and Bound
• Learning General Policies with Policy Gradient Methods