So the Bayesian belief is an expression for precisely everything the agent can possibly know about the world. Why not have the agent’s memory represent exactly that?
The usual reasons for a lossless compressor not compressing maximally are that it is expensive and slow to compress—and for many applications you need rapid random access to the data.
The usual reasons for a lossless compressor not compressing maximally are that it is expensive and slow to compress—and for many applications you need rapid random access to the data.