Machine Learning and Verification: No Room for Error

Varda Zilberman


Varda Zilberman always loved science. In high school she majored in physics and electronics, and later studied computer science and math at the Hebrew University, with the idea of pursuing a career in high tech. On campus, Varda felt that she’d found her place. As graduation neared, she deliberated whether to continue to a master’s degree or enter the workforce. She ended up working for a small tech company doing mapping for the Tel Aviv light rail. 

Yet Varda couldn’t ignore the lure of academia; after working for a year, she returned to the Hebrew University to begin her master’s degree. 

"I loved being a student and wanted to take my studies to the next step – conducting research. I wanted to push myself further and see what I could accomplish."

Now nearing the end of her first year, Varda has been working on machine learning in Dr. Guy Katz’s laboratory. More specifically, she studies the field of neural networks verification, i.e. verifying that a neural network satisfies some properties, and exploring ways to speed up the verification process – a computationally challenging task. Verification is crucial for many systems, such as self-driving car and airborne collision avoidance systems.

"I was always drawn to the more theoretical aspects of math, but there’s something immensely satisfying about practical research and getting immediate results, especially working in the field of verification, where there’s no room for error. I love my research and feel that I’m in the right place.