Finite Depth and Width Corrections to the Neural Tangent Kernel
This addresses the theoretical understanding of deep learning dynamics for researchers, revealing that deep and wide networks can learn features beyond shallow ones, though it is incremental in refining NTK theory.
The paper proves that in deep and wide ReLU networks, the neural tangent kernel (NTK) has exponential scaling in the depth-to-width ratio, making it non-deterministic even with infinite overparameterization, and shows that its first SGD update also scales exponentially, enabling data-dependent feature learning in the lazy training regime.
We prove the precise scaling, at finite depth and width, for the mean and variance of the neural tangent kernel (NTK) in a randomly initialized ReLU network. The standard deviation is exponential in the ratio of network depth to width. Thus, even in the limit of infinite overparameterization, the NTK is not deterministic if depth and width simultaneously tend to infinity. Moreover, we prove that for such deep and wide networks, the NTK has a non-trivial evolution during training by showing that the mean of its first SGD update is also exponential in the ratio of network depth to width. This is sharp contrast to the regime where depth is fixed and network width is very large. Our results suggest that, unlike relatively shallow and wide networks, deep and wide ReLU networks are capable of learning data-dependent features even in the so-called lazy training regime.