Basser Seminar Series

On Associative Classifiers and on Multiclass Imbalanced Data

Speaker: Professor Osmar Zaiane
University of Alberta, Canada

When: Wednesday 21 May, 2014, 4:00-5:00pm

Where: The University of Sydney, School of IT Building, SIT Lecture Theatre (Room 123), Level 1

Add seminar to my diary

Abstract

The talk will have two parts. We will present associative classifiers in the first part and will talk about dealing with imballanced training data in the second part (time premitting).

There are countless paradigms and strategies for devising a classifier. Associative classifiers use association rules as a model and are a relatively new approach to rule-based classification. While they are still not as effecive as other approaches, they have many advantages and certainly potential in many applications. We will briefly introduce associative classifiers, discuss their main three phases: rule generation, rule pruning and rule selection, and highlight the differences between the suggested strategies. We will also present our current research work targeting theses three individual phases to improve the effectiveness of such classifiers.

The performance of traditional classification algorithms can be limited on imbalanced datasets. In recent years, the imbalanced data learning problem has drawn significant interest. We propose a hybrid method that combines two ideas: diverse random subspace ensemble learning with evolutionary search, to improve the performance of a neural network classifier on multiclass imbalanced data.

Speaker's biography

Osmar R. Zaïane is a Professor in Computing Science at the University of Alberta, Canada, and Scientific Director of the Alberta Innovates Centre for Machinre Learning (AICML). Dr. Zaiane joined the University of Alberta in July of 1999. He obtained a Master's degree in Electronics at the University of Paris, France, in 1989 and a Master's degree in Computer Science at Laval University, Canada, in 1992. He obtained his Ph.D. from Simon Fraser University, Canada, in 1999 under the supervision of Dr. Jiawei Han. His Ph.D. thesis work focused on web mining and multimedia data mining. He has research interests in novel data mining algorithms, web mining, text mining, image mining, social network analysis, and information retrieval. He has published more than 150 papers in refereed international conferences and journals, and taught on all six continents. Osmar Zaiane is the Secretary-Treasurer of the ACM SIGKDD (Special Interest Group on Data Mining) and was the treasurer of ACM SIGHIT (Special Interest group on Health Informatics). He was the Associate Editor then Editor-inChief of the ACM SIGKDD Explorations from 2003 to 2010. He is also Associate Editor of the Knowledge and Information Systems, An International Journal, by Springer, and of the journal Data Mining and Knowledge Discovery by Springer, as well as the International Journal of Internet Technology and Secured Transactions He was the General co-Chair of the IEEE International Conference on Data Mining ICDM 2011. Osmar Zaiane received the ICDM Outstanding Service Award in 2009 and the 2010 ACM SIGKDD Service Award.