Network Intrusion Detection using RBF neural networks
- 0 Collaborators
Predicting network intrusion using Radial Basis Function SVM and artificial neural networks. ...learn more
Project status: Under Development
Overview / Usage
- To detect malicious network packets from dataset ICSX 2017 from UNB.
- To predict them using RBF neural networks and compare the performance of the algorithms with SVM kernels.
Use in production:-
To create routers that can predict malicious attacks before the intrusion. There is no device in the industry that does this, at the moment.
Methodology / Approach
Methodology:- Experimental design
- Labeled dataset is used to create a pattern of malicious network packets, using RBF NN.
- Evaluate the performance metrics such as accuracy, memory used etc.
- Compare the performance against the traditional SVM kernel with RBF.
- If kernel is found as a better performer, optimize the RBF NN to bring it within comparable range of RBF SVM.
Frameworks:-
TensorFlow in Python, Sci-kit learn
Techniques:-
ANN's such as RBF neural networks and shallow algorithms such as SVM with RBF kernels