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