The Landscape of Empirical Risk for Non-convex Losses

07/22/2016
by   Song Mei, et al.
0

Most high-dimensional estimation and prediction methods propose to minimize a cost function (empirical risk) that is written as a sum of losses associated to each data point. In this paper we focus on the case of non-convex losses, which is practically important but still poorly understood. Classical empirical process theory implies uniform convergence of the empirical risk to the population risk. While uniform convergence implies consistency of the resulting M-estimator, it does not ensure that the latter can be computed efficiently. In order to capture the complexity of computing M-estimators, we propose to study the landscape of the empirical risk, namely its stationary points and their properties. We establish uniform convergence of the gradient and Hessian of the empirical risk to their population counterparts, as soon as the number of samples becomes larger than the number of unknown parameters (modulo logarithmic factors). Consequently, good properties of the population risk can be carried to the empirical risk, and we can establish one-to-one correspondence of their stationary points. We demonstrate that in several problems such as non-convex binary classification, robust regression, and Gaussian mixture model, this result implies a complete characterization of the landscape of the empirical risk, and of the convergence properties of descent algorithms. We extend our analysis to the very high-dimensional setting in which the number of parameters exceeds the number of samples, and provide a characterization of the empirical risk landscape under a nearly information-theoretically minimal condition. Namely, if the number of samples exceeds the sparsity of the unknown parameters vector (modulo logarithmic factors), then a suitable uniform convergence result takes place. We apply this result to non-convex binary classification and robust regression in very high-dimension.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
07/11/2019

The Landscape of Non-convex Empirical Risk with Degenerate Population Risk

The landscape of empirical risk has been widely studied in a series of m...
research
06/05/2019

Robustness and Tractability for Non-convex M-estimators

We investigate two important properties of M-estimator, namely, robustne...
research
04/18/2023

High-dimensional Multi-class Classification with Presence-only Data

Classification with positive and unlabeled (PU) data frequently arises i...
research
05/19/2017

The Landscape of Deep Learning Algorithms

This paper studies the landscape of empirical risk of deep neural networ...
research
12/04/2020

Non-Asymptotic Analysis of Excess Risk via Empirical Risk Landscape

In this paper, we provide a unified analysis of the excess risk of the m...
research
12/04/2019

Landscape Complexity for the Empirical Risk of Generalized Linear Models

We present a method to obtain the average and the typical value of the n...
research
11/21/2016

Scalable Approximations for Generalized Linear Problems

In stochastic optimization, the population risk is generally approximate...

Please sign up or login with your details

Forgot password? Click here to reset