Sharp Recovery Thresholds of Tensor PCA Spectral Algorithms

Michael Jacob Feldman, David Donoho

Keywords: tensor PCA, spectral algorithms, random matrix theory

Abstract: Many applications seek to recover low-rank approximations of noisy tensor data. We consider several practical and effective matricization strategies which construct specific matrices from such tensors and then apply spectral methods; the strategies include tensor unfolding, partial tracing, power iteration, and recursive unfolding. We settle the behaviors of unfolding and partial tracing, identifying sharp thresholds in signal-to-noise ratio above which the signal is partially recovered. In particular, we extend previous results to a much larger class of tensor shapes where axis lengths may be different. For power iteration and recursive unfolding, we prove that under conditions where previous algorithms partially recovery the signal, these methods achieve (asymptotically) exact recovery. Our analysis deploys random matrix theory to obtain sharp thresholds which elude perturbation and concentration bounds. Specifically, we rely upon recent disproportionate random matrix results, which describe sequences of matrices with diverging aspect ratio.

Submission Number: 14297