Tue 25 Jun 2019 15:00 - 15:20 at 224AB - Reasoning and Optimizing ML Models Chair(s): Martin Maas

In recent years, the notion of local robustness (or robustness for short) has emerged as a desirable property of deep neural networks. Intuitively, robustness means that small perturbations to an input do not cause the network to perform misclassifications. In this paper, we present a novel algorithm for verifying robustness properties of neural networks. Our method synergistically combines gradient-based optimization methods for counterexample search with abstraction-based proof search to obtain a sound and (δ -)complete decision procedure. Our method also employs a data-driven approach to learn a verification policy that guides abstract interpretation during proof search. We have implemented the proposed approach in a tool called Charon and experimentally evaluated it on hundreds of benchmarks. Our experiments show that the proposed approach significantly outperforms three state-of-the-art tools, namely AI^2, Reluplex, and Reluval.

Conference Day
Tue 25 Jun

Displayed time zone: Tijuana, Baja California change

14:00 - 15:30
Reasoning and Optimizing ML ModelsPLDI Research Papers at 224AB
Chair(s): Martin MaasGoogle
14:00
20m
Talk
An Inductive Synthesis Framework for Verifiable Reinforcement Learning
PLDI Research Papers
He ZhuRutgers University, USA, Zikang XiongPurdue University, Stephen Magill, Suresh JagannathanPurdue University
Media Attached
14:20
20m
Talk
Programming Support for Autonomizing Software
PLDI Research Papers
Wen-Chuan LeePurdue University, Peng LiuPurdue University, Yingqi LiuPurdue University, USA, Shiqing MaPurdue University, USA, Xiangyu ZhangPurdue University
14:40
20m
Talk
Wootz: A Compiler-Based Framework for Fast CNN Pruning via Composability
PLDI Research Papers
Hui GuanNorth Carolina State University, Xipeng ShenNorth Carolina State University, Seung-Hwan LimOak Ridge National Laboratory, USA
Media Attached File Attached
15:00
20m
Talk
Optimization and Abstraction: A Synergistic Approach for Analyzing Neural Network Robustness
PLDI Research Papers
Greg AndersonUniversity of Texas at Austin, USA, Shankara PailoorUniversity of Texas at Austin, USA, Isil DilligUT Austin, Swarat ChaudhuriRice University
Media Attached