Yashoteja Prabhu

Yashoteja Prabhu 

Yashoteja Prabhu
Ph.D. Student
Department of Computer Science and Engineering
Indian Institute of Technology Delhi

I am a fifth year PhD student in the Computer Science and Engineering department of IIT Delhi. I am advised by Dr. Manik Varma. My primary research area is Extreme Multi-Label Classification, which is a sub-field of machine learning. I have designed and implemented several state-of-the-art extreme classification algorithms which can be useful in a wide range of web applications such as large-scale image classification, web-page tagging, product recommendation, computational advertising, and so on.

The Extreme Classification Repository is a one-stop resource center for extreme classification researchers. It provides datasets, code for baseline algorithms, benchmark results, information about evaluation metrics, and many more useful stuff.

Prior to joining the PhD program, I spent 2 wonderful years as a Research Assistant in the Machine Learning and Optimization group of Microsoft Research India. I graduated with B.Tech.(Hons) degree from CSE, IIT Bombay in 2011.

My CV is available here.


  • [10 October 2019] [New] Paper on Extreme Regressionaccepted for presentation at WSDM 2020. Preprint available here.

  • [22 December 2017] [New] Paper accepted for presentation at WWW 2018. The paper describes a new extreme classification algorithm, called Parabel, which achieves almost the same accuracy as the state-of-the-art 1-vs-All classifiers but is more than 1,000x and 10,000x faster during training and prediction respectively. The paper also demonstrates that Parabel significantly improves performance on the real world application of dynamic search advertising on Bing. Pdf available here.

  • [23 October 2017] [New] Paper accepted for presentation at WSDM 2018. The paper develops SwiftXML - a novel extreme classification algorithm for making accurate and scalable predictions on warm start tagging, recommendation and sponsored search advertising applications. Pdf available here.