1

Many researches have argued that Artificial Neural Networks (ANNs) can improve the performance of intrusion detection systems (IDS) when compared with traditional methods. However for ANN-based IDS, detection precision, especially for low-frequent attacks, and detection stability are still needed to be enhanced. A new approach is called FC-ANN, based on ANN and fuzzy clustering, to solve the problem and help IDS achieve higher detection rate, less false positive rate and stronger stability. The general procedure of FC-ANN is as follows: firstly fuzzy clustering technique is used to generate different training subsets. Subsequently, based on different training subsets, different ANN models are trained to formulate different base models. Finally, a meta-learner, fuzzy aggregation module, is employed to aggregate these results. Experimental results on the KDD CUP 1999 dataset show that the proposed new approach, FC-ANN, outperforms BPNN and other well-known methods such as decision tree, the naïve Bayes in terms of detection precision and detection stability.

Question:

Would it be possible to combine a Bayesian belief network/system with Fuzzy Clustering neural networks for intrusion detection?

Can anyone foresee any problems I may encounter? Your input would be most valuable.

Bojangles
  • 99,427
  • 50
  • 170
  • 208
G Gr
  • 6,030
  • 20
  • 91
  • 184
  • 1
    Which lit are you quoting above? Not sure how you want to combine Bayesian BN and Fuzzy Clustering neural networks. I don't think anyone can provide a useful answer when your question is at such a high level. – amit kumar Sep 11 '11 at 11:47
  • at first it was expert systems and NN's then fuzzy clustering/bayes decison tree's etc but no one is combing the qualities and tools of all and comparing them agaisnt there individual characteristics. The title for my research, Evaluation of Intelligent Methods within Network based Intrusion Detection Systems using Bayesian-Fuzzy Clustering neural networks. In this approach I want to test the validility of the strongest IDS performers against there individual qualitys and propose/develop a system/plugin for snort similar to spade. Input on some directon/bumps I might encounter would be good. – G Gr Sep 11 '11 at 12:05
  • I would not read too much into FC "winning" over (Back propagation NN) BPNN without carefully understanding the paper. Its your choice to combine a bayesian probabilistic approach with fuzzy techniques, but usually I would consider them as competitive approaches rather than cooperative approaches. – amit kumar Sep 11 '11 at 12:05
  • The simplest thing that could be tried is to combine two different (independent) "experts", in this case FC-NN and Bayesian BN, using some form of Boosting such as Adaboost. But I am not sure if your setting allows this. – amit kumar Sep 11 '11 at 12:13
  • 1
    Easier or not depends not only on its computational complexity but also on how well-developed the techniques are, and how much the user is familiar with them. Using a probabilistic approach, the uncertainty would be incorporated into a probability distribution (instead of a single value). – amit kumar Sep 11 '11 at 12:25
  • 2
    Another good forum for this kind of question is: metaoptimize.com – amit kumar Sep 11 '11 at 12:28
  • Garrith, glad to answer, but this is all I had to say. – amit kumar Sep 11 '11 at 12:46
  • 1
    Voting to close as off-topic. You might wanna try [CrossValidated](http://stats.stackexchange.com/) or [MetaOptimize](http://metaoptimize.com/qa/) – Amro Sep 11 '11 at 13:46

0 Answers0