Nicolas Duchateau

Reinforcement learning can now handle active feature selection in high-dimensional datasets by intelligently pruning the feature search space and regularizing decision sequences, outperforming existing methods in accuracy and policy complexity.