Conditional Generative Models are not Robust

06/04/2019
by   Ethan Fetaya, et al.
0

Class-conditional generative models are an increasingly popular approach to achieve robust classification. They are a natural choice to solve discriminative tasks in a robust manner as they jointly optimize for predictive performance and accurate modeling of the input distribution. In this work, we investigate robust classification with likelihood-based conditional generative models from a theoretical and practical perspective. Our theoretical result reveals that it is impossible to guarantee detectability of adversarial examples even for near-optimal generative classifiers. Experimentally, we show that naively trained conditional generative models have poor discriminative performance, making them unsuitable for classification. This is related to overlooked issues with training conditional generative models and we show methods to improve performance. Finally, we analyze the robustness of our proposed conditional generative models on MNIST and CIFAR10. While we are able to train robust models for MNIST, robustness completely breaks down on CIFAR10. This lack of robustness is related to various undesirable model properties maximum likelihood fails to penalize. Our results indicate that likelihood may fundamentally be at odds with robust classification on challenging problems.

READ FULL TEXT
research
02/19/2018

Are Generative Classifiers More Robust to Adversarial Attacks?

There is a rising interest in studying the robustness of deep neural net...
research
12/14/2022

Generative Robust Classification

Training adversarially robust discriminative (i.e., softmax) classifier ...
research
03/06/2017

Generative and Discriminative Text Classification with Recurrent Neural Networks

We empirically characterize the performance of discriminative and genera...
research
07/29/2020

Generative Classifiers as a Basis for Trustworthy Computer Vision

With the maturing of deep learning systems, trustworthiness is becoming ...
research
02/07/2020

Multimodal Controller for Generative Models

Class-conditional generative models are crucial tools for data generatio...
research
11/22/2017

Safer Classification by Synthesis

The discriminative approach to classification using deep neural networks...
research
06/19/2021

A Stealthy and Robust Fingerprinting Scheme for Generative Models

This paper presents a novel fingerprinting methodology for the Intellect...

Please sign up or login with your details

Forgot password? Click here to reset