Sparse Feature Selection Makes Batch Reinforcement Learning More Sample Efficient.
Botao HaoYaqi DuanTor LattimoreCsaba SzepesváriMengdi WangPublished in: ICML (2021)
Keyphrases
- feature selection
- reinforcement learning
- machine learning
- small sample
- batch mode
- multi agent systems
- high dimensional
- computationally efficient
- dimensionality reduction
- sparse data
- model free
- information gain
- function approximation
- high dimensionality
- computationally expensive
- mutual information
- data sets
- state space
- support vector
- training data
- neural network