Skip to main content

Showing 1–3 of 3 results for author: Mergny, P

Searching in archive stat. Search in all archives.
.
  1. arXiv:2506.02664  [pdf, other

    stat.ML cond-mat.dis-nn cs.LG

    Computational Thresholds in Multi-Modal Learning via the Spiked Matrix-Tensor Model

    Authors: Hugo Tabanelli, Pierre Mergny, Lenka Zdeborova, Florent Krzakala

    Abstract: We study the recovery of multiple high-dimensional signals from two noisy, correlated modalities: a spiked matrix and a spiked tensor sharing a common low-rank structure. This setting generalizes classical spiked matrix and tensor models, unveiling intricate interactions between inference channels and surprising algorithmic behaviors. Notably, while the spiked tensor model is typically intractable… ▽ More

    Submitted 3 June, 2025; originally announced June 2025.

  2. arXiv:2403.04234  [pdf, other

    stat.ML cs.LG

    Fundamental limits of Non-Linear Low-Rank Matrix Estimation

    Authors: Pierre Mergny, Justin Ko, Florent Krzakala, Lenka Zdeborová

    Abstract: We consider the task of estimating a low-rank matrix from non-linear and noisy observations. We prove a strong universality result showing that Bayes-optimal performances are characterized by an equivalent Gaussian model with an effective prior, whose parameters are entirely determined by an expansion of the non-linear function. In particular, we show that to reconstruct the signal accurately, one… ▽ More

    Submitted 7 March, 2024; originally announced March 2024.

    Comments: 42 pages, 2 figures

  3. arXiv:2403.03695  [pdf, other

    stat.ML cond-mat.dis-nn cs.LG math.PR math.ST

    Spectral Phase Transition and Optimal PCA in Block-Structured Spiked models

    Authors: Pierre Mergny, Justin Ko, Florent Krzakala

    Abstract: We discuss the inhomogeneous spiked Wigner model, a theoretical framework recently introduced to study structured noise in various learning scenarios, through the prism of random matrix theory, with a specific focus on its spectral properties. Our primary objective is to find an optimal spectral method and to extend the celebrated \cite{BBP} (BBP) phase transition criterion -- well-known in the ho… ▽ More

    Submitted 6 March, 2024; originally announced March 2024.

    Comments: 26 pages, 2 figures

    Journal ref: Proceedings of the 41st International Conference on Machine Learning, PMLR 235:35470-35491, 2024