An Information-Geometric Approach to Artificial Curiosity

ArXi:2504.06355v2 Announce Type: replace Learning in environments with sparse rewards remains a fundamental challenge in reinforcement learning. Artificial curiosity addresses this limitation through intrinsic rewards to guide exploration, however, the precise formulation of these rewards has remained elusive. Ideally, such rewards should depend on the agent's information about the environment, remaining agnostic to its representation -- an invariance central to information geometry.