Unpacking Reward Shaping: Understanding the Benefits of Reward Engineering on Sample Complexity.
Abhishek GuptaAldo PacchianoYuexiang ZhaiSham M. KakadeSergey LevinePublished in: NeurIPS (2022)
Keyphrases
- sample complexity
- reward shaping
- reinforcement learning
- theoretical analysis
- learning problems
- learning algorithm
- special case
- supervised learning
- active learning
- upper bound
- complex domains
- lower bound
- generalization error
- reinforcement learning algorithms
- markov decision problems
- sample size
- reward function
- function approximation
- data sets
- temporal difference
- support vector
- state space
- machine learning