I've been thinking a lot about how to apply FFTs to neural networks, so this is timely; thank you.

If you're looking for deeper explanations, I recommend "Practical Signal Processing": https://www.amazon.com/Practical-Signal-Processing-Mark-Owen...

Many of the chapters are here: https://battle.shawwn.com/sdb/books/PSP/ but buy the book if you can afford to.

In particular, Chapter 4's "untwisting" intuition is the best I've found: https://imgur.com/a/r7KvO60

Another way of phrasing it: You know that effect when a wheel is spinning so fast, it looks like it's going backwards? If it looks like it's standing still, there's a Fourier coefficient at that rotational frequency.

... Or at least, I think there is. It's hard to be precise with analogies.

I would be interested in knowing how it compares to Practical Signal Processing (https://www.amazon.com/Practical-Signal-Processing-Mark-Owen...)