Metadata-Version: 2.3
Name: airtk
Version: 0.1.2
Summary: Comprehensive and simple framework for conducting AI robustness experiments
Project-URL: Homepage, https://github.com/LAiSR-SK/AiRobustnessTestingKit-AiR-TK-
Project-URL: Issues, https://github.com/LAiSR-SK/AiRobustnessTestingKit-AiR-TK-/issues
Author-email: Deirdre Jost <jostdb@miamioh.edu>, Steven Chiacchira <chiaccsa@miamioh.edu>
License-File: AUTHORS.md
License-File: LICENSE.md
Classifier: Environment :: GPU :: NVIDIA CUDA :: 11.8
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: Microsoft :: Windows :: Windows 10
Classifier: Programming Language :: Python :: 3
Requires-Python: ==3.9
Requires-Dist: huggingface-hub>=0.24.0
Requires-Dist: networkx>=3.1
Requires-Dist: pytorch>=2.1.1
Requires-Dist: pyyaml>=6.0.0
Description-Content-Type: text/markdown

<!-- Logo needs to be touched up; add our name-->
<span align="center">

![Logo](asset/repo/image/logo.png)

[![Python Package](https://img.shields.io/pypi/pyversions/airtk?style=flat&logo=python&logoColor=green)](https://pypi.org/project/airtk/)
[![Huggingface](https://img.shields.io/badge/%F0%9F%A4%97%20HuggingFace-gray)](https://pypi.org/project/airtk/)
![Code Style](https://img.shields.io/badge/code_style-Ruff-orange)
![License](https://img.shields.io/github/license/LAiSR-SK/ImagePatriot)

</span>

As machine learning approaches to artificial intelligence continue to grow in popularity, the need for secure implementation and evaluation becomes increasingly paramount. This is of especially great concern in safety-critical applications such as object detection for self driving cars, monitoring nuclear power plants, and giving medical diagnoses. To this end we present a simple yet comprehensive interface for robust training and evaluation of PyTorch classifiers.

Although other solutions such as the [adversarial robustness toolbox](https://github.com/Trusted-AI/adversarial-robustness-toolbox) and [MAIR](https://github.com/Harry24k/MAIR) have provided solutions for this in the past, they are not as comprehensive in breadth of provided attacks and defenses.

# Installation
Our work is available via this repository and as a [PyPI package](https://pypi.org/project/airtk/).

## From PyPI (Recommended)
```bash
python3 -m pip install airtk
```

## From Repo Source (Not Recommended)
In order to install from here, you will need:
- The [Conda](https://www.anaconda.com/) environment manager.
- The [Git](https://www.git-scm.com/) version control system.

```bash
git clone https://github.com/LAiSR-SK/AiRobustnessTestingKit-AiR-TK-
```

```bash
conda env create -p .conda

conda activate ./.conda
```

# Contents
## Attacks
<!-- TODO: Aibak -->

## Defenses
You can import and use our defenses as shown:
```python
from torch import nn

from airtk.defense import TradesTraining

if __name__ == "__main__":
    # Initialize the training function
    training = TradesTraining(batch_size=512,
                              "cifar10",
                              "res101",
                              epochs=100,
                              lr=0.01,
                              seed=0,
                              model_dir="data/model/TRADES/",
                              save_freq=10)
                              
    # Run the specified training regime
    training()
```

We support the following defenses:
- Adversarial Distributional Training (ADT)
- Adversarial Distributional Training++ (ADTPP)
- Adversarial Weight Distribution ([ATAWP](https://arxiv.org/abs/2004.05884))
- Curriculum Adversarial Training (Currat)
- Federated Adversarial Training ([FAT](https://arxiv.org/pdf/2012.01791))
- Feature Scatter ([FS](https://arxiv.org/abs/1907.10764))
- Geometry Aware Instance Reweighted Adversarial Training ([GAIRAT](https://github.com/zjfheart/Geometry-aware-Instance-reweighted-Adversarial-Training))
- TRadeoff-inspired Adversarial DEfenses via Surrogate loss minimization ([TRADES](https://github.com/yaodongyu/TRADES))
- TRADES with Adversarial Weight Distribution (TRADESAWP)
- Various Attacks (VA)
- You Only Propogate Once ([YOPO](https://arxiv.org/abs/1905.00877))

Most of which can use the following keyword arguments:
| kwarg name    | use                                   |
|---------------|---------------------------------------|
| dataset_name  | name of the dataset to use            |
| model_name    | name of the model to use              |
| epochs        | number of epochs to train / test for  |
| batch_size    | size of training and testing batches  |
| eps           | size of image perturbations           |
| model_dir     | directory to save models to           |

## Pretrained Models
In order to expedite progress in the field of secure AI, we have provided the weights of our trained models on [huggingface](https://huggingface.co/LAiSR-SK). These can be loaded via `load_pretrained` and then or further augmented:

```python
import torch
from airtk.data import CIFAR100
from airtk.model import ResNet50
from torch.utils.data import DataLoader

if __name__ == "__main__":
    torch.set_default_device("cuda")

    # 1. Load the model
    model: ResNet50 = ResNet50.from_pretrained("LAiSR-SK/curriculum-at-cifar100-res50")
    
    # 2. Evaluate the model against CIFAR100
    testset: CIFAR100 = CIFAR100(root="data/", train=False, download=True)
    test_loader: DataLoader = DataLoader(testset, batch_szie = 256, shuffle=True)
    
    total: int = 0
    correct: int = 0
    for x, y in test_loader:
        logits = model(x)
        _, predicted = torch.max(logits, 1)

        total_correct += (predicted == y).sum().item()
        total += predicted.size[0]
        
    acc: float = 100 * correct / total

    print(f"Accuracy: {acc}")
```

## Cite Us
See [CITATION.cff](CITATION.cff) or the sidebar for details on how to cite our work.
