Approximating a deep reinforcement learning docking agent using linear model trees.
Vilde B. GjærumElla-Lovise H. RørvikAnastasios M. LekkasPublished in: CoRR (2022)
Keyphrases
- linear model
- reinforcement learning
- regression trees
- multi agent
- action selection
- regression model
- least squares
- linear models
- learning agent
- multi agent systems
- multiagent systems
- decision trees
- state space
- partially observable
- state abstraction
- nonlinear models
- linear transformation
- learning capabilities
- autonomous learning
- multiple agents
- exploration strategy
- intelligent agents
- multi agent environments
- agent receives
- agent learns
- function approximation
- learning agents
- mobile agents
- single agent
- linear transformations
- learning algorithm
- reward function
- autonomous agents
- state action
- software agents
- markov decision processes
- markov decision process
- semi parametric
- temporal difference
- model free
- dynamic environments
- machine learning
- action space
- function approximators
- optimal policy
- additive model