|
Activity Number:
|
241
- SLDS CPapers New
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, July 30, 2018 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Statistical Learning and Data Science
|
|
Abstract #330130
|
Presentation
|
|
Title:
|
Ensemble of Iterative Classifier Chains for Multi-Label Classification
|
|
Author(s):
|
Zhoushanyue He* and Matthias Schonlau
|
|
Companies:
|
University of Waterloo and University of Waterloo
|
|
Keywords:
|
Multi-label Classification;
Classification;
Statistical Learning;
Machine Learning
|
|
Abstract:
|
Multi-label classification is a classification problem in which an instance can be classified into more than one categories. Ensemble of Classifier Chains (ECC) is one of the benchmark methods for multi-label classification, in which classifier chains are applied in an ensemble framework. In this paper we extend the ECC approach by incorporating the probabilistic label relevance instead of 0/1 predictions and running classifier chains in iteration. The proposed Ensemble of Iterative Classifier Chains (EICC) and ECC are evaluated on real and simulated data. The empirical evaluation suggests that EICC inherits more information about the distribution of labels and produces better predictions when labels are unevenly distributed.
|
Authors who are presenting talks have a * after their name.
