UnsolvedMajor Unsolved Problem

Computational Threshold for Tensor PCA

Posed by Richard & Montanari (2014)

§ Problem Statement

Setup

Unsolved Problem

§ Discussion

Loading discussion…

§ Significance & Implications

§ Known Partial Results

§ References

[1]

A statistical model for tensor PCA

Emile Richard, Andrea Montanari (2014)

NeurIPS

📍 Section 1 (Introduction), in the “We next summarize our results” discussion contrasting “Ideal estimation” with “Tractable estimators: Unfolding/Power iteration” (computational-statistical gap), p. 2 (NeurIPS 2014 PDF)

[2]

Tensor principal component analysis via sum-of-squares proofs

Samuel Hopkins, Jonathan Shi, David Steurer (2015)

COLT

[3]

Notes on computational hardness of hypothesis testing: Predictions using the low-degree likelihood ratio

Dmitriy Kunisky, Alexander S. Wein, Afonso S. Bandeira (2019)

📍 Section 2.3 (Tensor PCA), Conjecture 2.4 (low-degree polynomial lower bound at the spectral threshold n^{-(k-1)/4}), p. 12.

§ Tags