Speeding Up Reinforcement Learning by Exploiting Causality in Reward Sequences.
Hongming LiJosé C. PríncipePublished in: IJCNN (2021)
Keyphrases
- reinforcement learning
- state space
- function approximation
- learning algorithm
- reinforcement learning algorithms
- reward function
- hidden markov models
- partially observable environments
- temporal difference
- model free
- learning problems
- mobile robot
- optimal control
- dynamic programming
- machine learning
- eligibility traces
- neural network
- markov decision processes
- causal relationships
- multi agent
- learning agent
- reward shaping