Cross Validation in Machine Learning

When you compare to the next two frames in the machine learning pilots, you might call that cross validation almost. It sounds like when you go further and further into the future with these double pendulums, do you see any does it still work? Do you see any differences between the complex models and the simple models in those cases? Yes, actually, that's a great question. We, when you use the sparse model that has only four variables, it can predict, it can generate future frames that look like it's much more stable in the long term. So we end up actually to do, to do it well, we use predictions at all scales. The simplicity leads to better predictions