aiqbench analysis
This arXiv paper is a theoretical statistics result on recovering a hidden vector from a noisy symmetric tensor, using random matrix theory rather than the i.i.d. Gaussian simplification common in prior work. Its main contribution is a sharp phase transition for the square unfolding algorithm at a signal strength of roughly n^((p-2)/4), along with conjectured tradeoffs between square and non-square unfoldings. For ML readers, the practical relevance is mainly to the theory of spectral methods for high-order data, not to any immediate system or benchmark.
A Random Matrix Analysis of Symmetric Tensor Unfolding
arXiv:2611.05137v1 Announce Type: cross Abstract: We study a fundamental problem in machine learning: partial recovery of a planted signal $\boldsymbol{v} \in \mathbb{S}^{n-1}$ from a spiked tensor $\boldsymbol{\mathsf{X}} = \lambda \, \boldsymbol{v}^{\otimes p} + \boldsymbol{\mathsf{Z}}$, where $\boldsymbol{\mathsf{Z}}$ has symmetric Gaussian entries of variance $\asymp n^{-1}$. While previous works replace $\boldsymbol{\mathsf{Z}}$ by an i.i.d. standard Gaussian tensor to circumvent technical difficulties caused by dependencies, this paper directly analyzes the symmetric model via random matrix theory. We show that the square unfolding algorithm is subject to a sharp phase transition for partial recovery at $\lambda \asymp n^{(p-2) / 4}$, validating an open conjecture from Montanari and Richard (2014). We get precise expressions for the limiting spectral distribution, deterministic equivalent, phase transition threshold and overlap. The main message is that unfolded symmetric tensors spectrally behave like standard Wigner matrices, up to simple transformations accounting for high-order symmetries. We consider extensions to multiple orthogonal spikes, recursive unfolding and non-square unfolding via disproportionate random matrix theory, leading to an interesting conjectured statistical-computational tradeoff regarding the choice of unfolding: square versus long.
Abstract:We study a fundamental problem in machine learning: partial recovery of a planted signal $\boldsymbol{v} \in \mathbb{S}^{n-1}$ from a spiked tensor $\boldsymbol{\mathsf{X}} = \lambda \, \boldsymbol{v}^{\otimes p} + \boldsymbol{\mathsf{Z}}$, where $\boldsymbol{\mathsf{Z}}$ has symmetric Gaussian entries of variance $\asymp n^{-1}$. While previous works replace $\boldsymbol{\mathsf{Z}}$ by an i.i.d. standard Gaussian tensor to circumvent technical difficulties caused by dependencies, this paper directly analyzes the symmetric model via random matrix theory. We show that the square unfolding algorithm is subject to a sharp phase transition for partial recovery at $\lambda \asymp n^{(p-2) / 4}$, validating an open conjecture from Montanari and Richard (2014). We get precise expressions for the limiting spectral distribution, deterministic equivalent, phase transition threshold and overlap. The main message is that unfolded symmetric tensors spectrally behave like standard Wigner matrices, up to simple transformations accounting for high-order symmetries. We consider extensions to multiple orthogonal spikes, recursive unfolding and non-square unfolding via disproportionate random matrix theory, leading to an interesting conjectured statistical-computational tradeoff regarding the choice of unfolding: square versus long.
Machine Learning (stat.ML); Machine Learning (cs.LG); Probability (math.PR)
arXiv:2611.05137 [stat.ML]
(or arXiv:2611.05137v1 [stat.ML] for this version)
https://doi.org/10.48550/arXiv.2611.05137 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 269 words
Related coverage
Folded Transport MCMC: Certifiable Quotient Posterior Computation for Symmetric Bayesian Models
arXiv · Jun 5, 2026
Two Clocks in Diffusion MLLMs: When Answers Stabilize Before Rationales Unfold
arXiv · Oct 2, 2026
You Talkin to Me?: A Network Analysis of Gendered Speaker-Addressee Patterns in Film Screenplays
arXiv · Jul 28, 2026
News
Quantum Computing Report · Jun 7, 2026