XAILib: Unified Library for Explainable AI

2023   ·   Project   Resources  

img XAILib: Unified Library for Explainable AI

An integrated Python library for Explainable AI with a unified interface for various explanation methods!

What is XAI-Lib?

XAI-Lib is an integrated Python library for Explainable AI (XAI) that provides a unified interface for various explanation methods. Developed as part of the Xai Project, XAI-Lib simplifies the process of explaining black-box models across different data types, making machine learning models more interpretable and transparent.

Key Features

XAI-Lib is designed to be modular, extensible, and easy to use:

  • Unified Interface: Simple, consistent API for multiple explanation methods
  • Multiple Data Types: Support for tabular, image, text, and time-series data
  • Extensible Architecture: Easy integration of new explanation methods
  • Model-Agnostic: Works with any black-box machine learning model
  • Well-Documented: Comprehensive documentation and examples

Supported Explanation Methods

For Tabular Data:

  • SHAP - SHapley Additive exPlanations
  • LIME - Local Interpretable Model-agnostic Explanations
  • Anchors - High-precision model-agnostic explanations
  • LORE - LOcal Rule-based Explanations

For Image Data:

  • GradCAM - Gradient-weighted Class Activation Mapping
  • LIME - Local Interpretable Model-agnostic Explanations
  • SHAP - SHapley Additive exPlanations
  • ABELE - Adversarial Black-box Explainer generating Latent Exemplars

For Text and Time Series Data:

  • Work in Progress - Coming soon!

Installation

The easiest way to install XAI-Lib is using pip:

pip install XAI-Library

For the latest development version, clone the repository and install in editable mode:

git clone https://github.com/kdd-lab/XAI-Lib.git
cd XAI-Lib
pip install -e .

Quick Start

Here’s a simple example of using LIME for tabular data explanation:

from xailib import Explainer

# Initialize your black-box model
# model = YourModel()

# Create an explainer
explainer = Explainer(model, method='lime')

# Generate explanation for a sample
explanation = explainer.explain(sample_data)

# Visualize the explanation
explainer.visualize(explanation)

For more examples and detailed usage, please check the examples/ directory.

Documentation

Complete documentation, tutorials, and API reference are available at:

Contributing

We welcome contributions! Please see our Contributing Guide for details on how to:

  • Report bugs and request features
  • Submit pull requests
  • Improve documentation
  • Add new explanation methods

Contact

For questions and support: