Learning Perception and Planning with Deep Active Inference
This work addresses a limitation in active inference research for AI and robotics by moving beyond predefined discrete state spaces, though it is incremental in integrating deep learning methods.
The paper tackled the problem of applying active inference to complex, continuous environments by using deep learning to learn state spaces and approximate probability distributions, enabling more flexible perception and planning.
Active inference is a process theory of the brain that states that all living organisms infer actions in order to minimize their (expected) free energy. However, current experiments are limited to predefined, often discrete, state spaces. In this paper we use recent advances in deep learning to learn the state space and approximate the necessary probability distributions to engage in active inference.