Fast Online Reinforcement Learning Control using State-Space Dimensionality Reduction.
Tomonori SadamotoAranya ChakraborttyJun-ichi ImuraPublished in: CoRR (2019)
Keyphrases
- state space
- reinforcement learning
- dimensionality reduction
- control problems
- reinforcement learning algorithms
- markov decision processes
- optimal control
- optimal policy
- adaptive control
- control strategies
- function approximation
- balancing exploration and exploitation
- dynamical systems
- heuristic search
- online learning
- continuous state spaces
- action space
- robot control
- dynamic programming
- state variables
- markov chain
- control system
- control policy
- high dimensional
- multi agent
- temporal difference
- markov decision problems
- action selection
- robotic systems
- robotic control
- partially observable
- nonlinear dimensionality reduction
- markov decision process
- data representation
- real time
- machine learning
- feature extraction
- model free
- particle filter
- data sets
- reward shaping
- computer vision
- control policies
- pattern recognition
- learning process
- low dimensional
- high dimensional data
- linear discriminant analysis
- high dimensionality
- planning problems
- reward function
- control method