Extensions of the Separate-and-Conquer Multilabel Rule Learner
Type of publication: | Mastersthesis |
Citation: | ba:salem |
Type: | Bachelor Thesis |
Year: | 2017 |
Month: | August |
School: | TU Darmstadt, Knowledge Engineering Group |
URL: | /lehre/arbeiten/bachelor/2017/Salem_Borhan-Youssef.pdf |
Abstract: | Multi-label classification is the task in Machine Learning to assign more than one label to an instance. Opposite to the single-label classification problem, where only a binary or a multi-class can be assigned to an instance, dependencies may exist between different labels in a multi-label problem. These dependencies can be used to improve the classification task and help to better understanding the multi-label dataset. A Separate-and-Conquer Multi-Label Rule Learner was proposed 2016 by Eneldo Loza MencĂa and Frederik Janssen, that learn multi-label dependency and use them in the classification task. In this work we made some extensions of the proposed algorithm and evaluate them. |
Userfields: | betreuer={ELM} |
Keywords: | |
Authors | |
Topics
|
|
|