The brain also appears to have dedicated network structures that are not trained, but constrained, i.e. programmed to one transformation, say, the equivalent of

The brain also appears to have dedicated network structures that are not trained, but constrained, i.e. programmed to one transformation, say, the equivalent of feeding both an image and it's edge enhanced version to the same DNN. Current approach of feeding raw bitmaps to DNN falls short of that and is very sensitive to training data[1] I remember an old paper I cannot find now about how to normalize image for NN processing in face recognition. Software extracted the face, centered it on a square and projected that square on a circle around the center to make face orientation irrelevant (hard to explain without images) Anyway, it is unfair to expect a DNN to perform vision recognition tasks from raw bi-dimensional image points. [1] http://www.i-programmer.info/news/105-artificial-intelligence/8064-the-deep-flaw-in-all-neural-networks.html http://www.i-programmer.info/news/105-artificial-intelligenc...