1

I would like to know, from the below classification algorithm which one is optimal(in terms of memory and processing power) and which classification algorithm used widely?

  • k-Nearest Neighbor(KNN) Algorithm
  • Naive Bayesian(NB) Algorithm
  • Concept Vector-based(CB) Algorithm
  • Singular Value Decomposition-based Algorithm
  • Hierarchical Algorithm
  • Combination Algorithm
  • SVM

I am looking for an optimal algorithm to classify network traffic.

Thank you!

-sathish

1
  • 4
    There is no 'best' solution without knowing the problem domain (what are you trying to classify in your network traffic?) Commented Dec 3, 2010 at 13:10

2 Answers 2

2

The statement that there is no single best classification method is one of the No Free Lunch theorems - http://www.no-free-lunch.org/ Since you mention processing time and memory requirements I note that Naive Bayes is pretty cheap and has been good enough to be used in practice.

At http://www.cs.waikato.ac.nz/ml/weka/ you can pick up a lot of ready-implemented methods and a framework that will allow you to try them out and compare them.

Very often what features you pick out to feed into your classification method is more important than what method you chose.

Sign up to request clarification or add additional context in comments.

Comments

1

IMO, SVM is best one and currently used in many application.

2 Comments

It's widely used but it's not the best choice for every problem. You should consider problem domain before.
@max taldykin, The OP asked for network traffic, and it can be used there, also other methods like KNN are old methods, and I think customization on each method make it useful for a given problem, not exactly the selected method, It's not a solution to the problems.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.