Events Calendar

CS Colloquium: Animashree Anandkumar (UC Irvine) - Guaranteed Non-convex Algorithms for Modern Machine Learning through Tensor Factorization

Speaker: Animashree Anandkumar, UC Irvine

Talk Title: Guaranteed Non-convex Algorithms for Modern Machine Learning through Tensor Factorization

Series: CS Colloquium

Abstract: This lecture satisfies requirements for CSCI 591: Computer Science Research Colloquium

Modern machine learning involves processing massive datasets of diverse varieties such as text, images, videos, biological data, and so on. Designing efficient algorithms which are guaranteed to learn in a fast and a scalable manner is one of grand challenges. Most machine learning tasks can be cast as optimization problems, but unfortunately a majority of them are NP-hard non-convex problems. I will provide broad guidelines for overcoming this hardness barrier by: (i) focusing on conditions which make learning tractable, (ii) replacing the given optimization objective with better behaved ones, and (iii) exploiting non-obvious connections that abound in learning problems.

I will demonstrate the above guidelines using concrete examples: (i) unsupervised learning of latent variable models and (ii) training multi-layer neural networks, through a new framework involving spectral decomposition of moment matrices and tensors. Tensors are rich structures that can encode higher order relationships in data. Despite being non-convex, tensor decomposition can be solved optimally using simple iterative algorithms under mild conditions. These positive results demonstrate that previous theory on computational hardness of learning is overly pessimistic, and that we need new theoretical tools to explain the recent empirical success of non-convex learning algorithms.

This meeting will be available to stream HERE. Please right-click, open in new tab for best results.

Biography: Anima Anandkumar is a faculty at the EECS Dept. at U.C.Irvine since August 2010. Her research interests are in the areas of large-scale machine learning, non-convex optimization and high-dimensional statistics. In particular, she has been spearheading the development and analysis of tensor algorithms for a variety of learning problems. She is the recipient of several awards such as the Alfred. P. Sloan Fellowship, Microsoft Faculty Fellowship, Google research award, ARO and AFOSR Young Investigator Awards, NSF CAREER Award, Early Career Excellence in Research Award at UCI, Best Thesis Award from the ACM SIGMETRICS society, IBM Fran Allen PhD fellowship, and best paper awards from the ACM SIGMETRICS and IEEE Signal Processing societies. She received her B.Tech in Electrical Engineering from IIT Madras in 2004 and her PhD from Cornell University in 2009. She was a postdoctoral researcher at MIT from 2009 to 2010, and a visiting faculty at Microsoft Research New England in 2012 and 2014.

Host: CS Department

Webcast: https://bluejeans.com/267704433

Location: Henry Salvatori Computer Science Center (SAL) - 101

WebCast Link: https://bluejeans.com/267704433

Audiences: Everyone Is Invited

Contact: Assistant to CS chair