(Original post: http://stackoverflow.com/questions/26149416/using-machine-learning-to-predict-the-collapse-stabilization-of-complex-system)
I should add that I've also come up since this stackoverflow post with various methods ranging from Windowing, to convolution, FFTs, using a variation of expansion meshing, and even a method that revolved around trying to perform a Monte Carlo tree search over a set of systematic samples (really overly complicated, didn't work).
I'm at a serious loss of ideas right now. This is a project that I'm doing with a 5 University students, and all of us are clueless on how to correctly approach this.
If anyone has an idea on how to approach this, please let me know.