Timezone: »

Learning to Utilize Shaping Rewards: A New Approach of Reward Shaping
Yujing Hu · Weixun Wang · Hangtian Jia · Yixiang Wang · Yingfeng Chen · Jianye Hao · Feng Wu · Changjie Fan

Thu Dec 10 09:00 AM -- 11:00 AM (PST) @ Poster Session 5 #1382

Reward shaping is an effective technique for incorporating domain knowledge into reinforcement learning (RL). Existing approaches such as potential-based reward shaping normally make full use of a given shaping reward function. However, since the transformation of human knowledge into numeric reward values is often imperfect due to reasons such as human cognitive bias, completely utilizing the shaping reward function may fail to improve the performance of RL algorithms. In this paper, we consider the problem of adaptively utilizing a given shaping reward function. We formulate the utilization of shaping rewards as a bi-level optimization problem, where the lower level is to optimize policy using the shaping rewards and the upper level is to optimize a parameterized shaping weight function for true reward maximization. We formally derive the gradient of the expected true reward with respect to the shaping weight function parameters and accordingly propose three learning algorithms based on different assumptions. Experiments in sparse-reward cartpole and MuJoCo environments show that our algorithms can fully exploit beneficial shaping rewards, and meanwhile ignore unbeneficial shaping rewards or even transform them into beneficial ones.

Author Information

Yujing Hu (NetEase Fuxi AI Lab)
Weixun Wang (Tianjin University)
Hangtian Jia (Netease Fuxi AI Lab)
Yixiang Wang (University of Science and Technology of China)
Yingfeng Chen (NetEase Fuxi AI Lab)
Jianye Hao (Tianjin University)
Feng Wu (University of Science and Technology of China)
Changjie Fan (NetEase Fuxi AI Lab)

More from the Same Authors