Mutation-Based Adversarial Attacks on Neural Text Detectors

Gongbo Liang, Jesus Guerrero, Izzat Alsmadi
2/11/2023
cs.CRcs.AI

No official code found. Best implementation: DaBr01/Adversarial-Attacks-on-Neural-Text-Detection (3 stars, last commit 2024-06-18).

Abstract

Neural text detectors aim to decide the characteristics that distinguish neural (machine-generated) from human texts. To challenge such detectors, adversarial attacks can alter the statistical characteristics of the generated text, making the detection task more and more difficult. Inspired by the advances of mutation analysis in software development and testing, in this paper, we propose character- and word-based mutation operators for generating adversarial samples to attack state-of-the-art natural text detectors. This falls under white-box adversarial attacks. In such attacks, attackers have access to the original text and create mutation instances based on this original text. The ultimate goal is to confuse machine learning models and classifiers and decrease their prediction accuracy.

View on arXivView PDF

Code Implementations(1)

DaBr01/Adversarial-Attacks-on-Neural-Text-DetectionCommunity70%

Code and data to evaluate adversarial attacks on neural text detectors

31Python2 years ago
adversarial-attacksai-detectionai-detectorgenerative-ai

Similar Papers

BERT-ATTACK: Adversarial Attack Against BERT Using BERT

86% match2 repos3.5k2020

Is BERT Really Robust? A Strong Baseline for Natural Language Attack on Text Classification and Entailment

86% match1 repo5322019

Rethinking Textual Adversarial Defense for Pre-trained Language Models

85% match2022

Adversarial Attacks on AI-Generated Text Detection Models: A Token Probability-Based Approach Using Embeddings

85% match2025

Generating Natural Language Attacks in a Hard Label Black Box Setting

85% match1 repo502020

Adversarial Sample Detection for Deep Neural Network through Model Mutation Testing

85% match1 repo262018

Cite this paper

@article{liang2023mutationbased,
  title  = {Mutation-Based Adversarial Attacks on Neural Text Detectors},
  author = {Gongbo Liang and Jesus Guerrero and Izzat Alsmadi},
  year   = {2023},
  eprint = {2302.05794},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2302.05794v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2302.05794.svg)](https://codeofpaper.com/paper/2302.05794)

Discussion