TAI DQ

(Click the image for demo video!)

Project Overview

TAI DQ focuses on two major themes: Data Quality (DQ) and Explainable AI (XAI). The project aims to automate workflows, reduce data processing costs, and enhance reliability.

Data Quality (DQ)

Explore the DQ Repository

• Implements parametric and non-parametric imputation methods to improve data usability.
• Evaluates imputation results using:
  • Differential Entropy
  • Canonical Relative Entropy
  • Correlation-based metrics

Explainable AI (XAI)

• Incorporates Active Learning for interactive and adaptive workflows.
• Ensures process stability and reduces parameters through Knowledge Distillation.
• Enhances model interpretability with:
  • SHAP (SHapley Additive exPlanations)
  • Counterfactual Explanations

These methods assist in informed and transparent decision-making.

Additional Resources

• Presentation PPT
• Poster

---

Website Preview

• Frontend: React, Tailwind CSS
• Backend: Django, PostgreSQL (containerized with Docker)

Setup Instructions

Prerequisites

• Docker Desktop
• Node.js & npm (for frontend)

Frontend Setup

Run the frontend on http://localhost:5173/:

cd frontend
npm install
npm run dev

Backend & Database Setup

The backend is containerized using Docker Compose. To set it up:

1. Start Docker Desktop.
2. Build and start the backend and database:

cd backend
docker compose up --build

Restarting Backend (Optional)

If Docker is already set up, restart the backend without rebuilding:

docker compose up

---

Notes

• CORS Policy: Configured using a Vite proxy.

Contact

For questions or issues, raise a GitHub issue or contact the project maintainers.

Enjoy exploring TAI DQ! 🚀