Interesting: there was a parallel branch of development of CNNs in ~1989 independent of LeCun. Neocognitron+Backprop https://drive.google.com/drive/folders/0B65v6Wo67Tk5RDdCV0wwWlRTVU0 …
-
.
@karpathy Here's another parallel stream https://papers.nips.cc/paper/20-an-artificial-neural-network-for-spatio-temporal-bipolar-patterns-application-to-phoneme-classification.pdf … (also check refs 13, 21). Not as detailed as the one you linked, though.
the notion of optical implementation of convnet-like systems is quite fascinating.
