2700, chemin de la Tour
Montréal (QC) Canada  H3T 1J4

Titre : Clustering Using Asymmetric Loss Function

Conférencier : Prof. Adel Mohammadpour

Date : jeudi 18 octobre 2018, de 11 h 30 à 12 h 20

Lieu : L-4812, pavillon Lassonde Polytechnique Montréal

Résumé :
There are many techniques to group the observations into clusters, which use the loss functions to measure the dissimilarities between all pairs of observations such as Manhattan, Euclidean, and Mahalanobis distances. A symmetric distance is useful for clustering when overestimating and underestimating of clusters are equally costly. That is, miss classifying object losses in two clusters have the same values. There is no common distance measure, which can be appropriate for all clustering applications. However, in some clustering problems, the use of an asymmetric loss function may be more suitable.

To make our point clear, we give the following example. Magic Gamma Telescope is used to depict high-energy gamma particles with the imaging technique in a ground-based aerial Cherenkov Gamma Telescope. It is split into two clusters, Gamma (signal), and Hadron (background). The simple clustering accuracy is not significant for this data, because clustering a background event as the signal is worse than clustering a signal event as the background.

In this talk, we introduce clustering methods with asymmetric Linear Exponential (LINEX) loss function. The theoretical and experimental justification for the proposed algorithm are presented. The introduced clustering method is compared with well-known algorithms through a few real datasets as well as some simulated datasets.

Bio :
Adel Mohammadpour is an associate professor at Tehran Polytechnic and visitor in Department of Mathematics and Statistics at McGill University.

Bienvenue à tous!

Consulté 323 fois   ·   Modifier