CISE Seminar: Anand D. Sarwate, The University of New Jersey, Rutgers

Flexible Tensor Decompositions for Learning and Optimization

Many measurements or signals are multidimensional, or tensor-valued. To fit this data in existing machine learning pipelines, this data is vectorized, causing a blowup in the dimensionality. An alternative approach is to use tensor decompositions to create more structured models that respect the multidimensional structure. In this work we propose a family such structured decompositions, which we call low separation rank (LSR) tensor models. In the talk I will relate these to classical decompositions and show how the LSR model can balance model complexity and performance in supervised and unsupervised learning. Time permitting, we will describe applications of these ideas in other machine learning problems.

Anand D. Sarwate Anand D. Sarwate is an Associate Professor in the Department of Electrical and Computer Engineeringat Rutgers, The State University of New Jersey. He is also a member of the graduate faculty in the Department of Computer Science and the Department of Statistics. Sarwate works on problems that involve probability, mathematical statistics, and optimization, with applications in information theory, communication, signal processing, and machine learning. He is particularly interested in how these things intersect in the context of distributed/decentralized systems with constraints like privacy, bandwidth, latency, power, and so on.

Faculty Host: Bobak Nazer

Student Host: Qian Wu