I like the Bayesian Surprise definition for this. It's not about predicting the exact next state of the world (or the next frame of the noisy TV) but about how

I like the Bayesian Surprise definition for this. It's not about predicting the exact next state of the world (or the next frame of the noisy TV) but about how much the next state changes your model of the world. https://papers.nips.cc/paper_files/paper/2005/hash/0172d289da48c48de8c5ebf3de9f7ee1-Abstract.html https://papers.nips.cc/paper_files/paper/2005/hash/0172d289d...