https://github.com/bogdan-kulynych/textfool

Plausible looking adversarial examples for text classification
https://github.com/bogdan-kulynych/textfool

adversarial-machine-learning adversarial-samples deep-learning machine-learning

Last synced: 2 months ago
JSON representation

Plausible looking adversarial examples for text classification

• Host: GitHub
• URL: https://github.com/bogdan-kulynych/textfool
• Owner: bogdan-kulynych
• License: mit
• Created: 2017-05-01T00:42:02.000Z (over 7 years ago)
• Default Branch: master
• Last Pushed: 2018-12-16T12:57:11.000Z (about 6 years ago)
• Last Synced: 2024-08-03T01:17:21.723Z (6 months ago)
• Topics: adversarial-machine-learning, adversarial-samples, deep-learning, machine-learning
• Language: Python
• Homepage:
• Size: 85 KB
• Stars: 92
• Watchers: 7
• Forks: 32
• Open Issues: 4
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• awesome-production-machine-learning - TextFool - kulynych/textfool.svg?style=social) - plausible looking adversarial examples for text generation. (Adversarial Robustness Libraries)
• Awesome-AIML-Data-Ops - TextFool - kulynych/textfool.svg?style=social) - plausible looking adversarial examples for text generation. (Adversarial Robustness Libraries)
• awesome-production-machine-learning - TextFool - kulynych/textfool.svg?style=social) - plausible looking adversarial examples for text generation. (Adversarial Robustness)

README

        
# Plausible looking adversarial examples for text classification

**This repository is deprecated. Check out [trickster](https://github.com/spring-epfl/trickster) library and the "[Evading classifiers in discrete domains with provable optimality guarantees](https://arxiv.org/abs/1810.10939)" paper for an evolution and a more mature version of this approach to generating adversarial examples.**

[![DOI](https://zenodo.org/badge/89886473.svg)](https://zenodo.org/badge/latestdoi/89886473)

This is a proof of concept aiming at producing "imperceptible" [adversarial examples](https://blog.openai.com/adversarial-example-research/) for text classifiers.

For instance, this are some adversarial examples produced by this code for a classifier of a tweet author's gender based on the tweet's text:

![Examples of adversarial examples](images/examples.png)

## Setup

### System

You need Python 3, and all system dependencies possibly required by

* Keras

* NLTK

* SpaCy

### Python

```

pip install -r requirements.txt

```

### NLP Data

* SpaCy English language model:

  ```

  python -m spacy download en

  ```

* NLTK datasets (a prompt will appear upon running `paraphrase.py`)

## Model

To train using default parameters simply run

```

python run_training.py

```

By default will check for the CSV data set at `./data/twitter_gender_data.csv`, and save the model weights to `./data/model.dat`.

Should attain about 66% accuracy on validation data set for gender recognition.

### Data

This model uses Kaggle [Twitter User Gender Classification](https://www.kaggle.com/crowdflower/twitter-user-gender-classification) data.

## Demo

To run the adversarial crafting script:

```

python run_demo.py

```

Success rate for crafting the adversarial example should be about 17%.

By default the script will write the crafted examples into `./data/adversarial_texts.csv`.

## Paraphrasing

This module is rather reusable, although not immensely useful for anything practical. It provides a function that "paraphrases" a text by replacing some words with their WordNet synonyms, sorting by GloVe similarity between the synonym and the original context window. Relies on SpaCy and NLTK.

Example of paraphrase:

![Paraphrase example](images/paraphrase_example.png)

## Citing notes

Please use [Zenodo link](https://zenodo.org/record/831638) to cite _textfool_. Not that this work is not published, and not peer-reviewed. _textfool_ has no relationship to _"Deep Text Classification Can be Fooled."_ by B. Liang, H. Li, M. Su, P. Bian, X. Li, and W. Shi.