Mix & Match Agent Curricula for Reinforcement Learning.
Wojciech Marian CzarneckiSiddhant M. JayakumarMax JaderbergLeonard HasencleverYee Whye TehNicolas HeessSimon OsinderoRazvan PascanuPublished in: ICML (2018)
Keyphrases
- reinforcement learning
- multi agent
- action selection
- multi agent systems
- intelligent agents
- learning agent
- multi agent environments
- state abstraction
- state action
- learning capabilities
- reinforcement learning agents
- multiagent systems
- partially observable
- function approximation
- learning agents
- markov decision process
- single agent
- exploration strategy
- agent receives
- multiple agents
- agent model
- state space
- dynamic environments
- mobile agents
- reward shaping
- autonomous learning
- reward function
- software agents
- machine learning
- autonomous agents
- agent architecture
- agent technology
- markov decision processes
- learning algorithm
- decision making
- multiagent learning
- model free
- agent learns
- multiagent reinforcement learning
- cooperative
- temporal difference
- artificial agents
- supervised learning
- cognitive agents
- reinforcement learning algorithms
- action space
- transfer learning