MailGuard AI

Overview

MailGuard AI is an intelligent spam detection system designed to identify and classify emails as either spam or not spam (ham). Utilizing a machine learning model with BERT-tiny for tokenization, the application helps users efficiently filter unwanted emails, ensuring only relevant communication is prioritized.

Technologies Used

• Language Model & Tokenization: BERT-tiny

• Machine Learning Library: ONNX Runtime (for running the model)

• Machine Learning Technologies:

  • TensorFlow
  • PyTorch
  • Transformers
  • Scikit-learn

• Backend Framework: FastAPI

• Frontend Technologies:

  • React
  • HTML
  • CSS
  • JavaScript

Model Performance

Below is the model's training performance, showing training loss per epoch and accuracy metrics:

Features

• Spam Detection: Uses a machine learning model to classify emails as "Spam" or "Ham".
• Interactive User Interface: A user-friendly interface to enter email content and view classification results.
• Search History: Stores previous search results for easy reference.
• Real-Time Inference: Fast processing of email content using ONNX Runtime for real-time feedback.
• Accuracy: The model achieved an accuracy of 0.92 during training.

Installation

1. Clone the repository:

   bash
   git clone https://github.com/Arhaans/MailGuard-AI.git
   

2. Navigate to the project directory:

   cd MailGuard-AI
   

3. Create a virtual environment:

   conda create --name mailguard-env
   

4. Activate the virtual environment:

   pip install -r requirements.txt
   

5. Install the required dependencies:

   pip install -r requirements.txt
   

Usage

1. Start the backend server:

• Run the following command to start the FastAPI server: bash

  uvicorn main:app --reload
  

• The backend will be accessible at http://127.0.0.1:8000.

2. Start the frontend:

• Ensure you are in the frontend directory (if separate).

• Run the frontend using a command like:

  npm start
  

• Open http://localhost:3000 in your web browser to interact with the MailGuard AI application.

3. Using the Application:

• Enter the email content in the provided input field.
• Click "Search" to classify the email as "Spam" or "Ham".
• View your search history and delete or clear entries as needed.

Future Work

• Advanced Features:
  • Implement user authentication for personalized histories.
  • Add options to save and export chat histories.
• Enhanced Model Capabilities:
  • Incorporate a more complex language model for even more accurate spam detection.
  • Enable customization for individual spam preferences.

Contributing

Contributions are welcome! Please open an issue or submit a pull request with any improvements or suggestions you may have.

License

This project is licensed under the MIT License. See the LICENSE file for more information.

Acknowledgements

• BERT-tiny for providing the foundation for tokenization and language processing.
• ONNX Runtime for enabling efficient model inference.
• React and FastAPI for the frontend and backend frameworks.