Implementation code for the paper "Generating Natural Language Adversarial Examples"

A generic library for crafting adversarial NLP examples - WIP

Adversarial examples for Seq2Seq model in NLP

EMNLP BlackBox NLP 2020: Searching for a Search Method: Benchmarking Search Algorithms for Generating NLP Adversarial Examples

NLP robust to adversarial examples.

EMNLP Findings 2020: Reevaluating Adversarial Examples in Natural Language

This code is the 4th place in the 3rd phase of the alibaba/tsinghua university security AI challenger program. Adversarial examples for NLP's Chinese insults.

对司法文本数据进行黑盒对抗攻击

Extending adversarial examples to natural language

Generating Natural Language Adversarial Examples for Chinese NLI model

Here we explore cross-domain transferability of adversarial examples in NLP.

The Adversarial Machine Learning TextFool Dataset is a curated JSONL-format dataset designed for evaluating and benchmarking NLP models against adversarial text attacks. It focuses on adversarial examples generated using the TextFooler technique, aiming to deceive models while preserving semantic meaning and human readability.

NLP course's final project

Reevaluating Adversarial Examples in Natural Language

Generating Natural Language Adversarial Examples for NLI model

Adversarial-Examples-for-Text-Classification-NLP

Applying Differential Privacy principles to achieve robustness against adversarial examples in neural NLP classification models.

StructAdv is an adversarial-example generation technique that aims to generate adversarial training examples for Q&A NLP models through presupplied token models.

Course project for COMS 6998-10 at Columbia University, generating black-box NLP adversarial examples based on sentence-level perturbations, saliency evaluation and genetic algorithm.