Clyde: A Deep Reinforcement Learning DOOM Playing Agent.
Dino Stephen RatcliffeSam DevlinUdo KruschwitzLuca CitiPublished in: AAAI Workshops (2017)
Keyphrases
- reinforcement learning
- multi agent
- action selection
- autonomous agents
- learning agent
- multi agent systems
- learning capabilities
- reward function
- multi agent environments
- state abstraction
- state action
- agent receives
- autonomous learning
- partially observable
- decision making
- multiagent systems
- function approximation
- intelligent agents
- single agent
- state space
- reward shaping
- reinforcement learning agents
- optimal policy
- markov decision process
- agent systems
- mobile agents
- dynamic programming
- agent learns
- multiple agents
- exploration strategy
- robocup soccer
- agent architecture
- action space
- reinforcement learning algorithms
- markov decision processes
- temporal difference
- software agents
- model free
- agent model
- agent technology
- imperfect information
- multi agent learning
- multi agent reinforcement learning
- machine learning