July 17, 2019

Verifying Properties of Gradient Boosting Model

  • Einziger G.
  • Goldstein M.
  • Saar Y.
  • Segall I.

Verifying properties of machine learning models is a critical challenge. Gradient Boosting is an important technique which allows weak learners to be combined into strong ones. This work is about verifying properties for Gradient Boosting models. Our main contribution is the encoding of such models as SMT expression which enables the verification of properties on such models. Specifically, we target the critical property of robustness to adversarial perturbations. Our evaluation show that for some datasets the models are much more robust than for others. We also show that it is easier to verify robustness for models with smaller depth rather than deep ones which are known to suffer from overfitting. To the best of our knowledge, our work is the first to verify properties of Gradient Boosting models.

View Original Article

Recent Publications

May 01, 2020

A Packaged 0.01-26-GHz Single-Chip SiGe Reflectometer for Two-Port Vector Network Analyzers

  • Chung H.
  • Ma Q.
  • Rebeiz G.
  • Sayginer M.

© 1963-2012 IEEE. This article presents a packaged SiGe BiCMOS reflectometer for 0.01-26-GHz two-port vector network analyzers (VNAs). The reflectometer chip is composed of a resistive bridge coupler and two wideband heterodyne receivers for coherent magnitude and phase detection. In addition, a high-linearity receiver channel is designed to accommodate 20 ...

August 01, 2019

Protecting photonic quantum states using topology

  • Blanco-Redondo A.

The use of topology to protect quantum information is well-known to the condensed-matter community and, indeed, topological quantum computing is a bursting field of research and one of the competing avenues to demonstrate that quantum computers can complete certain problems that classical computers cannot. In photonics, however, we are only ...

May 01, 2019

Digital networks at the nexus of productivity growth

  • Kamat S.
  • Prakash S.
  • Saniee I.
  • Weldon M.

This paper takes a fresh look at the debate over the relationship between digital technology and productivity. The argument of economic historian Robert J. Gordon is that digital technology will not lead to increases in productivity such as we saw in the last century, based on his analysis of the ...