Andrea Montanari A

Andrea Montanari A | JRNLClub

Research Statement

AI generated · 2d ago

Andrea Montanari's indexed work centers on message-passing algorithms for compressed sensing, developed with collaborators including David Donoho, which connects statistical physics methods to high-dimensional signal recovery and ranks among his most-cited contributions. Related work extends this line to the dynamics of message passing on dense graphs and to the noise-sensitivity phase transition in compressed sensing, while separate threads address confidence intervals and hypothesis testing for high-dimensional regression and matrix completion from a few entries, alongside earlier physics work on length and time scales in glassy systems. Beyond these, his portfolio includes middle-author contributions such as Gibbs states and the set of solutions of random constraint satisfaction problems and a mean-field view of the landscape of two-layer neural networks, the latter reflecting a sustained interest in high-dimensional ridgeless least-squares interpolation. No software tools, datasets, patents, or funded grants appear in the indexed record; the through-line is a statistical-physics-informed approach to high-dimensional inference and learning.

Most Cited Works

A mean field view of the landscape of two-layer neural networks

Song Mei, Andrea Montanari, Phan-Minh Nguyen

Proceedings of the National Academy of Sciences2018

504Citations

Generalization error of random feature and kernel methods: Hypercontractivity and kernel matrix concentration

Song Mei, Theodor Misiakiewicz, Andrea Montanari A

Applied and Computational Harmonic Analysis2021

51Citations

Research Areas

Sparse and Compressive Sensing TechniqueStatistical Methods and InferenceTheoretical and Computational PhysicsError Correcting Code TechniquesRandom Matrices and Applications

Affiliations & Education

Employment

2006–present· now
Stanford University(US)
Professor · Electrical Engineering and Statistics
2002–2010
Ecole Normale supérieure Département de Physique(FR)

Loading ORCID activity...

Andrea Montanari A

Sources