Spectra of the Conjugate Kernel and Neural Tangent Kernel for linear-width neural networks
This provides theoretical insights into kernel spectra for neural networks, which is incremental for understanding optimization and generalization in deep learning.
The paper tackles the eigenvalue distributions of Conjugate Kernel (CK) and Neural Tangent Kernel (NTK) for multi-layer neural networks in a linear-width asymptotic regime, showing they converge to deterministic limits described by Marcenko-Pastur maps and recursive equations, with validation on synthetic and CIFAR-10 data.
We study the eigenvalue distributions of the Conjugate Kernel and Neural Tangent Kernel associated to multi-layer feedforward neural networks. In an asymptotic regime where network width is increasing linearly in sample size, under random initialization of the weights, and for input samples satisfying a notion of approximate pairwise orthogonality, we show that the eigenvalue distributions of the CK and NTK converge to deterministic limits. The limit for the CK is described by iterating the Marcenko-Pastur map across the hidden layers. The limit for the NTK is equivalent to that of a linear combination of the CK matrices across layers, and may be described by recursive fixed-point equations that extend this Marcenko-Pastur map. We demonstrate the agreement of these asymptotic predictions with the observed spectra for both synthetic and CIFAR-10 training data, and we perform a small simulation to investigate the evolutions of these spectra over training.