There was a chess neural network AlphaZero that taught itself how to play chess better than any chess program that has been created before in just 4 hours. I'm not asking to have it explained to me how they work because I already know that the answer is going to be too confusing for me. I'm asking whether anyone has a deep understand of how they work and knows everything they need to know about the generality of how they work. I think it's useful to know if anyone has a deep understanding on the inner workings of noural networks. I read the question How do neural networks manage to do regression? and skimmed the link in its answer http://neuralnetworksanddeeplearning.com/chap4.html without bothering to try and study it hard to understand it. That answer just gives me the information that it can compute any function. It probably meant it can compute any function within the time it can be computed by some algorithm up to a multiplicative factor. However that information is not enough for me. I want to know if all the relevant details that have some reason to unambiguously stand out of how neural networks work are known by anyone. The information of whether anyone has that kind of deep understanding of them is probably useful for the world to know.