Network Intrusion Detection using Light Weight ML Ensemble Method

Codes for the paper entitled "Optimization of Predictive Performance of Intrusion Detection System Using Hybrid Ensemble Model for Secure Systems"

Following datasets were used in this study.

• The UNSW-NB15 Dataset
• NSL-KDD dataset
• CSE-CIC-IDS2018

Following algorithms were studied individually and an ensemble of RF, SVM and MLP was used to develop a cost-effective and accurate model for Intrusion Detection.

• Random Forest
• Decision Tree
• kNN
• SVM
• MLP
• DNN
• CNN (4 Conv1D layers)
• LSTM (3 LSTM layers)
• RNN (3 RNN layers)

Evaluation Methods

• Accuracy
• Precision
• Recall
• F1-Score

Machine Specifications used for experimentation

• HP 840 G2 laptop
• Intel core i5 processor (5th generation)
• 64 bit Windows 10 operating system
• 16 GB RAM

Note

These codes are obtained by downloading .py files from original jupyter notebooks. While reproducing the results make sure you run these codes in a separate Jupyter Notebook.

Update

Codes for deep learning algorithms will be uploaded soon.