Extreme multi-label learning : A large scale classification approach in machine learning. Issue 4 (19th May 2019)

• Record Type: Journal Article
• Authors: Prajapati, Purvi; Thakkar, Amit
• Is Part Of: Journal of information & optimization sciences. Volume 40:Issue 4(2019)
• Page Start: 983
• Page End: 1001
• Publication Date: 2019-05-19
• Subjects: 68U35; Machine Learning -- Multi-Label Classification (MLC) -- Recommendation System; Electronic data processing -- Periodicals; Information science -- Periodicals; Mathematical optimization -- Periodicals; 519.6
• DOI: 10.1080/02522667.2019.1598000
• Languages: English
• ISSNs: 0252-2667

Abstract

Abstract: In digital world, the amount of data is growing exponentially in day to day life. It is difficult to analyze and extract knowledge from large amount of data with millions of categories in Big Data environment. Therefore, it is challenging problem to develop model that classify large volume of documents available on Internet. However, Multi-Label Classification approach is used to classify data with multiple categories or labels but it is inefficient way to deal with millions of categories. Hence Extreme Multi-Label Classification approach is used to overcome this limitation by selecting subset of labels for the new instance from millions of labels. Recently Extreme Multi-Label Classification has attracted research attention in different application areas like document categorization in Wikipedia, people identification in social networking, gene prediction in bio-informatics etc. Extreme Multi-Label Classification is also opened up new challenge to reformulate existing machine learning problems like ranking, tagging and recommendation. This survey paper focuses on approaches and reviewing current research challenges on eXtreme Multi Label Classification. Also discussed state-of-the-art algorithms to handle eXtreme Multi-Label Classification Problem.