Without doubt, the best mathematical explanation for inner workings of neural networks. Great work, Sergen!

You could try Dynamic Time Warping (DTW) method (https://rpubs.com/esobolewska/dtw-time-series). It's useful for quantifying similarity between pairs of time sequences . The DTW measure will be small for two series that exhibit similar peaks (or troughs) even if they are slightly displaced in time.

This here is the MVP. Bookmarking for future reference.

Great warmup before reading the actual paper. Thanks!