Streaming Kernel PCA with Õ(√(n)) Random Features
share
We study the statistical and computational aspects of kernel principal component analysis using random Fourier features and show that under mild assumptions, O(√(n) n) features suffices to achieve O(1/ϵ^2) sample complexity. Furthermore, we give a memory efficient streaming algorithm based on classical Oja's algorithm that achieves this rate.
READ FULL TEXT