You are cordially invited to attend the following seminar on Wed. 10 February 2016, 2:30 pm, Room 22-119

 

Due to their theoretical basis and their high accuracy in practice, SVMs are considered one of the best “off-the-shelf” supervised learning algorithm. Primarily developed for binary classification, various techniques have been used to extend SVMs for multi-class classification. State-of-art SVMs are highly optimized for large scale and online learning and one of the techniques used is called LaSVM. LaSVM is a SVM-based classifier that uses an optimization scheme inspired by online algorithms in favor to faster reach accuracy levels comparable to batch SVM solvers. In these seminar we present LaSVM with different SVM-based techniques for multi-class classification.

 

Armin Kobilica is Master student at ICS department. He obtained his Bachelor degree from University Utara Malaysia. Before joining KFUPM, he has worked in telecommunication industry as a developer and has experience in academia as a teaching assistant and lab tutor. Currently, he is interested in applications of Statistical Learning Theory and Bioinformatics.