Distributional Temporal Difference Learning

Concept

A method in reinforcement learning that represents future rewards not as a single expected value, but as a distribution of possible outcomes, leading to richer representation learning and accelerated performance.

Mentioned in 1 video