WAGIC - WPF for MAGIC

WAGIC is a WPF-based graphical user interface for interacting with MAGIC - Detecting Advanced Persistent Threats via Masked Graph Representation Learning

Overview

WAGIC provides a user-friendly graphical interface for:

• Training MAGIC models on different datasets
• Evaluating and assessing model performance
• Automatically generating and displaying visualization charts
• Real-time monitoring of training/evaluation progress
• Managing trained models

System Requirements

Frontend (WAGIC)

• Windows 10/11
• .NET 8.0 or higher
• Visual Studio 2022 (for development)

Backend (MAGIC)

• Python 3.8+
• PyTorch
• Flask
• NumPy, Pandas, Matplotlib
• Other dependencies (see requirements.txt in MAGIC repo)

Installation and Setup

Step 1: Clone and setup MAGIC backend

# Clone MAGIC repository
git clone https://github.com/WanThinnn/MAGIC.git
cd MAGIC

# Install Python dependencies
pip install -r requirements.txt

# Run Flask API server
python app.py

Backend will run at: http://localhost:5000

Step 2: Build and run WAGIC

# Clone WAGIC repository
git clone https://github.com/WanThinnn/WAGIC.git
cd WAGIC

# Build project
dotnet build

# Run application
dotnet run --project WAGIC

Or open WAGIC.sln in Visual Studio and run.

Supported Datasets

WAGIC supports the following datasets:

• cadets - DARPA TC Cadets dataset
• fivedirections - Five Directions dataset
• streamspot - StreamSpot dataset
• theia - DARPA TC Theia dataset
• trace - DARPA TC Trace dataset
• wget - Wget dataset

Usage Guide

1. Training Model

1. Select dataset from "Dataset Selection" dropdown
2. Click "Start Training"
3. Monitor progress via progress bar and log
4. Wait for completion - model will be saved automatically

2. Evaluation Model

1. Select dataset that has a trained model
2. Click "Start Evaluation"
3. View results displayed on the interface:
   • AUC, F1, Precision, Recall
   • Confusion Matrix (TN, FN, TP, FP)
   • Test AUC with standard deviation

3. Visualization

• Automatic: Charts are generated automatically after evaluation
• Manual: Click "Create Chart" button anytime
• Result Window: Separate window displays charts with zoom/pan functionality

API Endpoints

WAGIC communicates with MAGIC backend through the following APIs:

Endpoint	Method	Description
/api/datasets	GET	Get list of available datasets
/api/models	GET	Get list of trained models
/api/train	POST	Start training process
/api/train/status	GET	Get training status
/api/eval	POST	Start evaluation process
/api/eval/status	GET	Get evaluation status
/api/visualize	POST	Generate visualization charts
/api/latest-result-image	GET	Get latest result image

Project Structure

WAGIC/
├── WAGIC/
│   ├── MainWindow.xaml      # Main interface
│   ├── MainWindow.xaml.cs   # Main logic
│   ├── result_windows.xaml  # Result display interface
│   ├── result_windows.xaml.cs # Result window logic
│   ├── App.xaml            # App configuration
│   └── WAGIC.csproj        # Project file
├── WAGIC.sln               # Solution file
└── README.md               # This file

Troubleshooting

API Connection Error

Error connecting to API: http://localhost:5000

Solution: Ensure MAGIC backend is running on port 5000

Invalid Dataset Error

Invalid dataset. Choose from: [cadets, fivedirections, ...]

Solution: Select dataset from available dropdown options

Model Not Trained Error

No model found for dataset X. Please train first!

Solution: Run training for that dataset before evaluation

Unicode Error in Subprocess

UnicodeDecodeError: 'charmap' codec can't decode byte

Solution: Fixed in backend with encoding='utf-8'

Collaborator

• Lại Quan Thiên - WanThinnn
• Mai Nguyễn Nam Phương - Mai Nguyen Nam Phuong - Cyber Security
• Trần Thế Hữu Phúc - tranthehuuphuc
• Hồ Diệp Huy - hohuyy

Acknowledgments

• MAGIC Framework - MAGIC Repository
• DARPA TC Program - For providing the datasets used
• PyTorch Team - Main machine learning framework
• Microsoft WPF Team - GUI framework

---

Star this repository if you find it helpful!