Boosting Verified Training for Robust Image Classifications via Abstraction

03/21/2023
by   Zhaodi Zhang, et al.
0

This paper proposes a novel, abstraction-based, certified training method for robust image classifiers. Via abstraction, all perturbed images are mapped into intervals before feeding into neural networks for training. By training on intervals, all the perturbed images that are mapped to the same interval are classified as the same label, rendering the variance of training sets to be small and the loss landscape of the models to be smooth. Consequently, our approach significantly improves the robustness of trained models. For the abstraction, our training method also enables a sound and complete black-box verification approach, which is orthogonal and scalable to arbitrary types of neural networks regardless of their sizes and architectures. We evaluate our method on a wide range of benchmarks in different scales. The experimental results show that our method outperforms state of the art by (i) reducing the verified errors of trained models up to 95.64 602.50x speedup; and (iii) scaling up to larger models with up to 138 million trainable parameters. The demo is available at https://github.com/zhangzhaodi233/ABSCERT.git.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
06/01/2022

On the Perils of Cascading Robust Classifiers

Ensembling certifiably robust neural networks has been shown to be a pro...
research
06/29/2022

Adversarial Ensemble Training by Jointly Learning Label Dependencies and Member Models

Training an ensemble of different sub-models has empirically proven to b...
research
04/22/2019

Optimization + Abstraction: A Synergistic Approach for Analyzing Neural Network Robustness

In recent years, the notion of local robustness (or robustness for short...
research
04/22/2019

Optimization and Abstraction: A Synergistic Approach for Analyzing Neural Network Robustness

In recent years, the notion of local robustness (or robustness for short...
research
04/01/2021

Towards Evaluating and Training Verifiably Robust Neural Networks

Recent works have shown that interval bound propagation (IBP) can be use...
research
08/22/2017

Scheduling Constraint Based Abstraction Refinement for Multi-Threaded Program Verification

Bounded model checking is among the most efficient techniques for the au...

Please sign up or login with your details

Forgot password? Click here to reset