Machine Learning and Science Forum — Mutual Information Estimation for Tensor Network Machine Learning

ML&Sci Forum

December 7, 2020
11:00am to 12:00pm
Virtual Participation

Machine Learning and Science Forum — Mutual Information Estimation for Tensor Network Machine Learning
Date: Monday, December 7, 2020
Time: 11:00 AM - 12:00 PM Pacific Time
Location: Participate remotely using this Zoom link 

Mutual Information Estimation for Tensor Network Machine Learning

Speaker: Ian Convy, Graduate Student, Whaley Group, UC Berkeley 
Abstract: Recently, there has been a small but growing effort to develop machine learning architectures from large tensor contraction schemes known as tensor networks. These networks have long been used as variational ansatzes in quantum many-body physics due to their ability to reproduce the different entanglement scaling patterns commonly found in ground-state systems. If the entanglement structure of a given system is known, then the connectivity of the tensor network ansatz can be tailored to match it. When applying tensor networks to machine learning, it seems natural to consider whether such a correspondence could also exist between the network and the correlation structure present in a given dataset. In this talk I will first provide an overview of tensor network methods in machine learning and explain their original motivation in many-body physics. I will finish by describing some of my efforts to characterize the correlation scaling of image datasets in an analogy to the entanglement scaling analysis performed on physical systems, which was met with mixed success.

The BIDS Machine Learning and Science Forum (formerly the Berkeley Statistics and Machine Learning Forum) was launched in Spring 2018 and currently meets biweekly (during the spring and fall semesters) to discuss current applications across a wide variety of research domains in the physical sciences and beyond. Hosted by BIDS Faculty Affiliate Uroš Seljak (professor of Physics at UC Berkeley) and BIDS Research Affiliate Ben Nachman (Physicist and Staff Scientist at Lawrence Berkeley National Laboratory), these active sessions bring together domain scientists, statisticians, and computer scientists who are either developing state-of-the-art methods or are interested in applying these methods in their research. All interested members of the UC Berkeley and Berkeley Lab communities are welcome and encouraged to attend.  To receive email notifications about upcoming meetings, or to request more information, please contact


Ian Convy

Graduate Student, Whaley Group, UC Berkeley

Ian Convy a third year PhD student in Professor Birgitta Whaley's group studying applications of machine learning to quantum computing and applications of quantum computing to machine learning.