>The ideal compression (from an intelligence point of view) is both information efficient and maximally composable This leads the compression to be overfitted

>The ideal compression (from an intelligence point of view) is both information efficient and maximally composable This leads the compression to be overfitted to the learning environment. Not that our intelligence is entirely immune to that either