Unconventional Formats of Background Knowledge from Human Teacher in Reward Shaping.
Syed Ali RazaMary-Anne WilliamsPublished in: HRI (Companion) (2017)
Keyphrases
- background knowledge
- reward shaping
- human teacher
- reinforcement learning
- imitation learning
- inductive logic programming
- reinforcement learning algorithms
- domain knowledge
- human subjects
- learning process
- logic programs
- learning mechanism
- metadata
- state space
- prior knowledge
- knowledge base
- domain theory
- markov decision processes
- reward function
- model free
- function approximation
- feature selection
- learning algorithm
- semantic information
- neural network