Universal Adversarial Perturbation for Text Classification

10/10/2019
by   Hang Gao, et al.
0

Given a state-of-the-art deep neural network text classifier, we show the existence of a universal and very small perturbation vector (in the embedding space) that causes natural text to be misclassified with high probability. Unlike images on which a single fixed-size adversarial perturbation can be found, text is of variable length, so we define the "universality" as "token-agnostic", where a single perturbation is applied to each token, resulting in different perturbations of flexible sizes at the sequence level. We propose an algorithm to compute universal adversarial perturbations, and show that the state-of-the-art deep neural networks are highly vulnerable to them, even though they keep the neighborhood of tokens mostly preserved. We also show how to use these adversarial perturbations to generate adversarial text samples. The surprising existence of universal "token-agnostic" adversarial perturbations may reveal important properties of a text classifier.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
10/26/2016

Universal adversarial perturbations

Given a state-of-the-art deep neural network classifier, we show the exi...
research
12/01/2019

A Method for Computing Class-wise Universal Adversarial Perturbations

We present an algorithm for computing class-specific universal adversari...
research
11/19/2021

Meta Adversarial Perturbations

A plethora of attack methods have been proposed to generate adversarial ...
research
11/16/2017

Defense against Universal Adversarial Perturbations

Recent advances in Deep Learning show the existence of image-agnostic qu...
research
04/30/2020

TextAT: Adversarial Training for Natural Language Understanding with Token-Level Perturbation

Adversarial training is effective in improving the robustness of neural ...
research
11/18/2020

Adversarial Turing Patterns from Cellular Automata

State-of-the-art deep classifiers are intriguingly vulnerable to univers...
research
02/20/2018

On Lyapunov exponents and adversarial perturbation

In this paper, we would like to disseminate a serendipitous discovery in...

Please sign up or login with your details

Forgot password? Click here to reset