Deep Learning-Based Network Intrusion Detection System (NIDS)

# Deep Learning-Based Network Intrusion Detection System (NIDS)

Designed and developed a Deep Learning-based Network Intrusion Detection System (NIDS) for unsupervised anomaly detection in network traffic. The system combines a Bidirectional LSTM Encoder with a Context-Conditioned WGAN-GP architecture to identify zero-day attacks and previously unseen cyber threats without requiring attack labels during training.

## Key Features

* Bidirectional LSTM Encoder for temporal sequence modeling.

* Context-Conditioned Wasserstein GAN with Gradient Penalty (WGAN-GP).

* Unsupervised anomaly detection trained exclusively on normal network traffic.

* Hybrid anomaly scoring using:

* Reconstruction Error (MSE)

* Discriminator Confidence

* Feature-wise Deviation

* Sliding-window sequence generation for temporal learning.

* Percentile-based threshold calibration for adaptive anomaly detection.

## Technologies

Python • TensorFlow • Keras • Deep Learning • BiLSTM • WGAN-GP • Machine Learning • Cybersecurity • Network Security • NSL-KDD • NumPy • Pandas • Scikit-learn • CustomTkinter

## Performance Evaluation

The model was evaluated using multiple performance metrics, including:

* Precision

* Recall

* F1-Score

* ROC-AUC

* Confusion Matrix

The proposed framework demonstrated strong separation between normal and anomalous network traffic while maintaining robust generalization against previously unseen attack patterns.

## Desktop Dashboard

Developed an interactive desktop application using CustomTkinter that provides:

* Real-time anomaly monitoring

* Dynamic alert threshold visualization

* Live packet-processing statistics

* Security event logging

* Explainable AI (XAI) feature importance analysis

## My Contributions

* Data preprocessing and feature engineering

* Deep learning model development

* WGAN-GP implementation

* Sequence generation and threshold calibration

* Model evaluation and performance analysis

* Desktop GUI development

* System integration and testing

This project was developed collaboratively with my team as part of an advanced cybersecurity and deep learning research project.