PixL2R: Guiding Reinforcement Learning Using Natural Language by Mapping Pixels to Rewards.
Prasoon GoyalScott NiekumRaymond J. MooneyPublished in: CoRR (2020)
Keyphrases
- reinforcement learning
- natural language
- markov decision processes
- machine learning
- reinforcement learning algorithms
- knowledge representation
- state space
- function approximation
- multi agent
- model free
- reward shaping
- input image
- natural language processing
- reward function
- natural language interface
- learning algorithm
- optimal policy
- information extraction
- image pixels
- semantic analysis
- learning process
- supervised learning
- pixel values
- optimal control
- natural language generation
- function approximators
- average reward