Neural Discrete Abstraction of High-Dimensional Spaces: A Case Study In Reinforcement Learning.
Petros GiannakopoulosAggelos PikrakisYannis CotronisPublished in: EUSIPCO (2020)
Keyphrases
- high dimensional spaces
- reinforcement learning
- nearest neighbor
- high dimensional data
- high dimensional
- fitted q iteration
- euclidean distance
- high dimensions
- dimensionality reduction
- state abstraction
- low dimensional
- network architecture
- dimensional data
- continuous state
- neural network
- space partitioning
- high dimensionality
- multi variate
- high dimensional datasets
- approximate nearest neighbor
- high level
- real world
- databases
- learning algorithm
- optimal policy
- state space
- markov decision processes
- data mining
- sensory inputs
- data points
- dynamic programming
- machine learning
- feature space
- pattern recognition