Song Mei

is this you? claim profile

∙ 0 followers
Stanford University

When Do Neural Networks Outperform Kernel Methods?

For a certain scaling of the initialization of stochastic gradient desce...
06/24/2020 ∙ by Behrooz Ghorbani, et al. ∙ 93 ∙ share

read it
The generalization error of random features regression: Precise asymptotics and double descent curve

Deep learning methods operate in regimes that defy the traditional stati...
08/14/2019 ∙ by Song Mei, et al. ∙ 1 ∙ share

read it
Limitations of Lazy Training of Two-layers Neural Networks

We study the supervised learning problem under either of the following t...
06/21/2019 ∙ by Behrooz Ghorbani, et al. ∙ 0 ∙ share

read it
Linearized two-layers neural networks in high dimension

We consider the problem of learning an unknown function f_ on the d-dime...
04/27/2019 ∙ by Behrooz Ghorbani, et al. ∙ 0 ∙ share

read it
Proximal algorithms for constrained composite optimization, with applications to solving low-rank SDPs

We study a family of (potentially non-convex) constrained optimization p...
03/01/2019 ∙ by Yu Bai, et al. ∙ 0 ∙ share

read it
Mean-field theory of two-layers neural networks: dimension-free bounds and kernel limit

We consider learning two layer neural networks using stochastic gradient...
02/16/2019 ∙ by Song Mei, et al. ∙ 0 ∙ share

read it
TAP free energy, spin glasses, and variational inference

We consider the Sherrington-Kirkpatrick model of spin glasses with ferro...
08/23/2018 ∙ by Zhou Fan, et al. ∙ 0 ∙ share

read it
A Mean Field View of the Landscape of Two-Layers Neural Networks

Multi-layer neural networks are among the most powerful models in machin...
04/18/2018 ∙ by Song Mei, et al. ∙ 0 ∙ share

read it
The landscape of the spiked tensor model

We consider the problem of estimating a large rank-one tensor u^⊗ k∈( R...
11/15/2017 ∙ by Gérard Ben Arous, et al. ∙ 0 ∙ share

read it
Solving SDPs for synchronization and MaxCut problems via the Grothendieck inequality

A number of statistical estimation problems can be addressed by semidefi...
03/25/2017 ∙ by Song Mei, et al. ∙ 0 ∙ share

read it
The Landscape of Empirical Risk for Non-convex Losses

Most high-dimensional estimation and prediction methods propose to minim...
07/22/2016 ∙ by Song Mei, et al. ∙ 0 ∙ share

read it

Song Mei

Featured Co-authors

Andrea Montanari

John Duchi

Yu Bai

Zhou Fan

Gérard Ben Arous

Behrooz Ghorbani

Theodor Misiakiewicz

Mihai Nica

Roberto I. Oliveira

Phan-Minh Nguyen

When Do Neural Networks Outperform Kernel Methods?

The generalization error of random features regression: Precise asymptotics and double descent curve

Limitations of Lazy Training of Two-layers Neural Networks

Linearized two-layers neural networks in high dimension

Proximal algorithms for constrained composite optimization, with applications to solving low-rank SDPs

Mean-field theory of two-layers neural networks: dimension-free bounds and kernel limit

TAP free energy, spin glasses, and variational inference

A Mean Field View of the Landscape of Two-Layers Neural Networks

The landscape of the spiked tensor model

Solving SDPs for synchronization and MaxCut problems via the Grothendieck inequality

The Landscape of Empirical Risk for Non-convex Losses