So I’d like to search for the best convolution operation that works best for entropy-encoding waveforms from a certain physics experiment. As you might imagine, the search space is very large, and I would need a machine learning approach to find a good minima. I have both lots of data and a supercomputer at my disposal.

I was thinking of using a similar approach to AlphaTensor (is there an open source alternative to it?), but I’m not well versed with ML algorithms, so any suggestions would be appreciated.

I have searched for non ML approaches, but think one would work best.