M-Tahar Kechadi, ProfessorSchool of Computer Science, University College Dublin (UCD), Ireland.
Speech Title: Dimensionality Reduction with Bivariate and Multivariate Copulas
Abstract: Modelling high dimensional datasets is a challenging task, as often the datasets contain noise and redundant dimensions, which lead to misspecification of the models and poor results of the analysis. Many statistical methods have been proposed to deal with this issue, but most of them present very high computational complexity and / or produce poor results. In this presentation, I discuss two approaches based on copula to detect inter-correlations between dimensions. The two approaches both use copulas. The first approach is the direct application of copulas to describe and model the inter-correlation (also called dependence) between any two dimensions - bivariate analysis. The second uses multivariate copulas to model dependences between a number of dimensions. While the two approaches are similar, they use different algorithms to model the inter-correlations between dimensions. After explaining their concepts, I will show how to use them to detect redundant dimensions and then compare their computational complexity and the quality of their results.
Biography: M-Tahar Kechadi is professor in School of Computer Science, University College Dublin (UCD), Ireland. He was awarded PhD and a Masters degree in Computer Science from University of Lille 1, France. He is currently Principal Investigator in the INSIGHT Centre for Data Analytics and CONSUS, one of the biggest projects in Precision Agriculture. He is in the editorial board of the Journal of Future Generation of Computer Systems. He is full member at CERN and a visiting professor at Fuzhou University, Fujian, China. The core and central focus of my research in the last decade is how to manage and analyse data quickly and efficiently. Nowadays we live in digital world, we produce more data than we can analyse and exploit. This “big data” will continue to grow at rapid pace, will underpin new waves of innovation in nearly every sector of the economy worldwide, and will reshape the way we build and use computers (hardware and software). Currently, my research interests are primary in i) Big Data Analytics and its applications to real-world applications. ii) Big Data Applications: Digital Healthcare and digital agriculture (Precision Agriculture). iii)Distributed Mining techniques and models and their execution environments and applications. vi) Cloud/Grid computing and services for supporting data access, management, and mining processes, v) Digital Forensics and cybercrime investigations.