Net Intrusion Detection Save
Network intrusion detection with Machine Learning (Deep Learning) experiment : 1d-cnn, softmax, neural networks, convolution
Project README
Deep Learning based network intrusion detection in PyTorch
Net intrusion detection experiment for Final Project of DeepLearning class at Inha University.
Dataset homepage: https://www.unb.ca/cic/datasets/ids-2017.html
Contributions
- Correct Evaluation Metric
- Adressing data imblance
- Benchmark results for different ML models
- Running code for training/evaluating
Accompanying slides
https://docs.google.com/presentation/d/1Rjj1vF0hv8vSJWeDxk23nE4A4w3fv8tBdvsyIBpWTdU/edit?usp=sharing
Model Performance using K-Fold Cross-Validation
| Classifier | 5-Fold Balanced Accuracy |
|---|---|
| Content Linear Softmax | 76.27 |
| Neural Network with 3 dense layer | 85.73 |
| Neural Network with 5 dense layer | 85.63 |
| 1D-CNN with 2conv 1fc layer | 87.13 |
| CNN with 5conv layer | 87.16 |
| Random Forest | 80.09 |
Softmax
Please run the Softmax.ipynb
NN
Please run the NN.ipynb There are two NN architectures:
- 'nn3' - 3 layers
- 'nn5' - 5 layers
1D-CNN
Please run the CNN.ipynb There are two 1D-CNN architectures:
- 'cnn2' - 2 conv layers
- 'cnn5' - 5 conv layers
Open Source Agenda is not affiliated with "Net Intrusion Detection" Project. README Source: Jumabek/net_intrusion_detection
Stars
35
Open Issues
0
Last Commit
3 months ago
Repository
Open Source Agenda Badge
