Beyond NTK- A Mean-Field Analysis of Neural Networks with Polynomial Width, Samples, and Time

Beyond NTK- A Mean-Field Analysis of Neural Networks with Polynomial Width, Samples, and Time

Explore a comprehensive lecture on advanced neural network analysis, focusing on the mean-field approach for networks with polynomial width, samples, and time. Delve into Tengyu Ma's insights from Stanford University as he presents cutting-edge research that goes beyond Neural Tangent Kernel (NTK) theory. Gain a deeper understanding of optimization and algorithm design in the context of neural networks, examining how these factors interplay to affect network performance and capabilities.