Monthly Archives: June 2011

Interesting multiple feature combination using SVM

In pattern recognition we may describe the same object using different features. Often we find out quickly that each feature type has its advantages and disadvantages. The logical question that arises then is how to build a classifier that profits … Continue reading

Posted in Uncategorized | Leave a comment

Right way to do cross-validation

In statistical learning, it is important how one utilizes his data for learning. Especially, how it is used for model parameter estimation. I will not give example how it is often done (and wrong) but instead a correct way immediately. … Continue reading

Posted in Uncategorized | Leave a comment

Notes on practical Multiple Kernel Learning

There are several issues to deal with when trying to get MKL working on your data. First is the choice of the toolbox. Currently, the two most efficient and easy to use are : SimpleMKL [LINK] SMO-MKL [LINK] Where the … Continue reading

Posted in Uncategorized | Leave a comment