Andrea Montanari: The landscape of some statistical problems

Speaker: Andrea Montanari, Stanford University

Date: Wednesday, April 19, 2017

Time: 4:30 PM to 5:30 PM Note: all times are in the Eastern Time Zone

Refreshments: 4:00 PM

Public: Yes

Location: refreshments in the G5 Lounge - Seminar in Room 2-190

Event Type:

Room Description:

Host: Ankur Moitra

Contact: Deborah Goodwin, 617.324.7303,

Relevant URL:

Speaker URL: None

Speaker Photo:

Reminders to:,

Reminder Subject: TALK: Andrea Montanari: The landscape of some statistical problems

Abstract: 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 (each example).
Studying the landscape of the empirical risk is useful to understand the computational complexity
of these statistical problems. I will discuss some generic features that can be used to prove that the
global minimizer can be computed efficiently even if the loss in non-convex.
A different mechanism arises in some rank-constrained semidefinite programming problems. In this case,
optimization algorithms can only be guaranteed to produce an (approximate) local optimum, but all local
optima are close in value to the global optimum.

Finally I will contrast these with problems in which the effects of non-convexity are more dramatic.

[Based on joint work with Yu Bai, Song Mei, Theodor Misiakiewicz and Roberto Oliveira]

Research Areas:

Impact Areas:

See other events that are part of the Theory of Computation (TOC) 2017.

Created by Deborah Goodwin Email at Monday, April 10, 2017 at 10:55 AM.