Ask HN: Are there any research using chaos theory to study Deep neural network?
I just know and reading this book: Emergence: From Chaos To Order (https://www.amazon.com/Emergence-Chaos-Order-Helix-Books/dp/0738201421),
but it just studied simple shadow neural network,
Are there any research using chaos theory to study Deep neural network? A lot of it you can check here https://scholar.google.com/scholar?hl=en&as_sdt=0%2C5&q=chao... or https://scholar.google.com/scholar?hl=en&as_sdt=0%2C5&q=chao... Also try to search for chaos in reinforcement learning. Hope it will help Chaotic systems are iterative, that is, you keep data through the same function over and over again. As usually used, neural networks run in a single pass, you get data in and get data out. If you iterated a deep neural network you'd probably find chaos and other dynamic phenomena. But if you just run it once, nope.