Manik Varma

Principal Researcher, Microsoft Research India
Adjunct Professor of Computer Science, IIT Delhi
<manik@microsoft.com>

I am a Principal Researcher at Microsoft Research India and an Adjunct Professor of computer science at the Indian Institute of Technology (IIT) Delhi. My research interests lie in the areas of machine learning, computational advertising and computer vision. Classifiers that I have developed have been deployed on millions of devices around the world and have protected them from viruses and malware. My algorithms are also generating millions of dollars on the Bing search engine (up to sign ambiguity). In 2013, John Langford and I coined the term extreme classification and found that we had inadvertently started a new area in machine learning. Today, by happenstance, extreme classification is thriving in both academia and industry with my classifiers being used in various Microsoft products as well as in the wider tech sector. I recently proclaimed 2 KB (RAM) ought to be enough for everybody prompting the international media to cover my research and compare me to Bill Gates (unfair, I'm more handsome!). I have been awarded the Microsoft Gold Star award, the Microsoft Achievement award, the WSDM Best Paper award, won the PASCAL VOC Object Detection Challenge and stood first in chicken chess tournaments and Pepsi drinking competitions. I have served as an area chair/senior PC member for machine learning, artificial intelligence and computer vision conferences such as AAAI, CVPR, ICCV, ICML, IJCAI and NIPS and am serving as an associate editor of the IEEE TPAMI journal. I am also a failed physicist (BSc St. Stephen's College, David Raja Ram Prize), theoretician (BA Oxford, Rhodes Scholar), engineer (DPhil Oxford, University Scholar), mathematician (MSRI Berkeley, Post-doctoral Fellow) and astronomer (Visiting Miller Professor, UC Berkeley).

Research

I am interested in the following research areas

• Machine learning: Machine learning for the Internet of Things, extreme classification, recommender systems, multi-label learning, resource-efficient machine learning, deep learning, supervised learning.
• Computer vision: Image search, object recognition, text recognition, texture classification.
• Computational advertising: Text ads, dynamic search ads, product ads, native ads, ad ranking, relevance and selection, query recommendation.

Joining my group: I am looking for full time PhD students at IIT Delhi and Research Fellows at Microsoft Research India to work with me on research problems in supervised machine learning, extreme classification, recommender systems, deep learning and resource constrained machine learning for the Internet of Things. Please e-mail your CV to me directly, in addition to formally applying to IIT/Microsoft's programmes, even though I might not be able to personally respond to your mail in most cases. IIT offers PhD Fellowships in collaboration with Microsoft and many other labs such as Facebook, Google, IBM and TCS. Please look at the web pages of the CSE Department and The School of IT for more details and other funding opportunities.

Projects: Unfortunately, I am unable to supervise projects of students outside IIT Delhi. If you are an external student and would like to work with me then the best way would be to join IIT Delhi's PhD programmes or apply for a Research Fellowship at MSR India. I apologize in advance for my inability to respond to e-mails about summer or winter projects, thesis supervision external to IIT Delhi, etc.

Internships: If you are a PhD student looking to do an internship with me then please e-mail me directly. I have only one or two internship slots and competition is stiff so please apply early. Unfortunately, I get more e-mails about internships than I can deal with, so please accept my apologies for not being able to respond to your mail if you are not a PhD student.

Professional Activities

• Programme Co-chair: ICVGIP 2014


• Area Chair/Senior PC Member: IJCAI 2019, AAAI 2018, IJCAI 2016 - AI on the Web, ICML 2016, ICML 2015, CVPR 2014, ACCV 2014, ICCV 2013, ICVGIP 2012, NIPS 2011, ICVGIP 2010


• Workshop Co-chair: Dagstuhl Extreme Classification Seminars, Extreme Classification 2017, TinyML 2017, Extreme Classification 2016, Extreme Classification 2015, RecTech 2015, The MSRI Machine Learning Summer School, Extreme Classification 2013, WebVision 2012, The Mysore Park Computer Vision Workshop, The MSRI Computer Vision & Graphics Shindig, The Winter School on Machine Learning and Computer Vision


• Keynotes and Selected Invited Talks: BigMine@KDD 2018, XMLC@WWW 2018, ACM India 2018 Summit, CODS+COMAD 2018, NASSCOM's Annual Tech Conference 2017, DSI@KDD 2016, DICTA 2015, Big Targets@ECML/PKDD 2015, X@ICML 2015, Budgeted ML@ICML 2015, LSOLDM 2014, ISI Kolkata ML Unit Founder's Day Lecture 2014, BDA 2013, IIIT Delhi Institute Lecture 2013, ICVGIP 2012, MPMLW 2012


• Media: A podcast about extreme classification, Living Science Media - Interview, Living Science Media - Talk, A blog about training a hundred million classifers, A blog about a fast, accurate, stable & tiny gated RNN, A blog about AI's big leap to tiny devices


• Teaching: 2009/CSL864   2013/CSV884   2016/SIV895   2018/COV878   2019/COV878  

Code

• The Extreme Classification Repository: Multi-label Datasets and Code
• Microsoft's Edge Machine Learning Library - FastGRNN, EMIRNN, Bonsai & ProtoNN: Running efficient ML models on severely resource constrained IoT microcontrollers & Arduino boards


• Slice: An extreme classifier for low-dimensional (deep learning) features that can scale to 100 million labels
• Parabel: Partitioned Label Trees for Extreme Classification
• SwiftXML: Extreme Classification with Label Features for Warm-start Tagging, Ranking and Recommendation
• PfastreXML: Propensity scored re-ranked FastXML
• SLEEC: Sparse Local Embeddings for Extreme multi-label Classification
• FastXML: A Fast, Accurate and Stable Tree Classifier for eXtreme Multi-label Learning
• M3L: Efficient Max Margin Multi-label Learning


• LDKL: Local Deep Kernel Learning for efficient non-linear prediction
• SPG-GMKL: Spectral Projected Gradient descent based optimization for Generalized Multiple Kernel Learning
• SMO-MKL: SMO based optimization for p-norm regularized Multiple Kernel Learning
• GMKL: Generalized Multiple Kernel Learning

Datasets

• The Extreme Classification Repository: Multi-label Datasets and Code
• The Chars74K Dataset: Character Recognition in Natural Images and an associated Julia tutorial and Kaggle competition (I have no idea how "Google" crept into the dataset name)
• CUReT: The Cropped Columbia-Utrecht Texture Classification Dataset & Associated Filterbanks

Current PhD Students

• Kunal Dahiya
• Himanshu Jain
• Pranshu Jain
• Yashoteja Prabhu

Some Former Students

• Kush Bhatia (PhD student at Berkeley, MSRI Research Fellow)
• Rahul Garg (Senior Research Scientist at Google, PhD UW, BTech IIT Delhi)
• Varun Gulshan (Research Scientist at Google, DPhil Oxford, BTech IIT Delhi)
• Saurabh Gupta (Assistant Professor at UIUC, PhD Berkeley, BTech IIT Delhi)
• Bharath Hariharan (Assistant Professor at Cornell, PhD Berkeley, BTech IIT Delhi)
• Ashesh Jain (Head of Perception at Lyft, PhD Cornell, BTech IIT Delhi
• Suraj Jain (Applied Scientist at Microsoft, MSRI Research Fellow)
• Anil Kag (PhD student at BU, MSRI Research Fellow)
• Ashish Kumar (PhD student at Berkeley, MSRI Research Fellow)
• Adarsh Prasad (PhD student at CMU, BTech IIT Delhi)
• Manish Singh (PhD student at MIT, BTech IIT Delhi)
• Deepak Vasisht (PhD student at MIT, BTech IIT Delhi)

Publications

• S. Bengio, K. Dembczynski, T. Joachims, M. Kloft and M. Varma. Extreme Classification (Dagstuhl Seminar 18291). Dagstuhl Reports, 8(7):62--80, 2019.
  Bibtex source | Abstract | Download in pdf format



• H. Jain, V. Balasubramanian, B. Chunduri and M. Varma. Slice: Scalable linear extreme classifiers trained on 100 million labels for Related Searches. In Proceedings of the ACM International Conference on Web Search and Data Mining, Melbourne, Australia, February 2019.
  Bibtex source | Abstract | Download in pdf format | Code | Extreme Classification Repository | Blog | Podcast



• A. Kusupati, M. Singh, K. Bhatia, A. Kumar, P. Jain and M. Varma. FastGRNN: A fast, accurate, stable and tiny kilobyte sized gated Recurrent Neural Network. In Advances in Neural Information Processing Systems, Montreal, Canada, December 2018.
  Bibtex source | Abstract | Download in pdf format | Code | Short Video | Blog



• Y. Prabhu, A. Kag, S. Harsola, R. Agrawal and M. Varma. Parabel: Partitioned label trees for extreme classification with application to Dynamic Search Advertising. In Proceedings of the ACM International World Wide Web Conference, Lyon, France, April 2018.
  Bibtex source | Abstract | Download in pdf format | Code | Extreme Classification Repository | Podcast



• Y. Prabhu, A. Kag, S. Gopinath, K. Dahia, S. Harsola, R. Agrawal and M. Varma. Extreme multi-label learning with label features for warm-start tagging, ranking and recommendation. In Proceedings of the ACM International Conference on Web Search and Data Mining, Los Angeles, California, February 2018.
  Bibtex source | Abstract | Download in pdf format | Code | Extreme Classification Repository



• A. Kumar, S. Goyal and M. Varma. Resource-efficient machine learning in 2 KB RAM for the Internet of Things. In Proceedings of the International Conference on Machine Learning, Sydney, Australia, August 2017.
  Bibtex source | Abstract | Download in pdf format | Code | Blog



• C. Gupta, A. Suggala, A. Gupta, H. Simhadri, B. Paranjape, A. Kumar, S. Goyal, R. Udupa, M. Varma and P. Jain. ProtoNN: Compressed and accurate kNN for resource-scarce devices. In Proceedings of the International Conference on Machine Learning, Sydney, Australia, August 2017.
  Bibtex source | Abstract | Download in pdf format | Code



• H. Jain, Y. Prabhu and M. Varma. Extreme multi-label loss functions for recommendation, tagging, ranking & other missing label applications. In Proceedings of the ACM SIGKDD Conference on Knowledge Discovery and Data Mining, San Francisco, California, August 2016.
  Bibtex source | Abstract | Download in pdf format | Code | Extreme Classification Repository | Talk



• K. Bhatia, H. Jain, P. Kar, M. Varma and P. Jain. Sparse local embeddings for extreme multi-label classification. In Advances in Neural Information Processing Systems, Montreal, Canada, December 2015.
  Bibtex source | Abstract | Download in pdf format | Code | Extreme Classification Repository



• Y. Prabhu and M. Varma. FastXML: A fast, accurate and stable tree-classifier for extreme multi-label learning. In Proceedings of the ACM SIGKDD Conference on Knowledge Discovery and Data Mining, New York, New York, August 2014.
  Bibtex source | Abstract | Download in pdf format | Code | Extreme Classification Repository | Slides | Talk on Extreme Classification & FastXML



• D. Vasisht, A. Damianou, M. Varma and A. Kapoor. Active learning for sparse Bayesian multi-label classification. In Proceedings of the ACM SIGKDD Conference on Knowledge Discovery and Data Mining, New York, New York, August 2014.
  Bibtex source | Abstract | Download in pdf format



• P. Jawanpuria, M. Varma and J. Saketha Nath. On p-norm path following in multiple kernel learning for non-linear feature selection. In Proceedings of the International Conference on Machine Learning, Beijing, China, June 2014.
  Bibtex source | Abstract | Download in pdf format



• C. Jose, P. Goyal, P. Aggrwal and M. Varma. Local deep kernel learning for efficient non-linear SVM prediction. In Proceedings of the International Conference on Machine Learning, Atlanta, Georgia, June 2013.
  Bibtex source | Abstract | Download in pdf format | Code | Slides



• R. Agrawal, A. Gupta, Y. Prabhu and M. Varma. Multi-label learning with millions of labels: Recommending advertiser bid phrases for web pages. In Proceedings of the International World Wide Web Conference, Rio de Janeiro, Brazil, May 2013.
  Bibtex source | Abstract | Download in pdf format | Slides



• A. Jain, S. V. N. Vishwanathan and M. Varma. SPG-GMKL: Generalized multiple kernel learning with a million kernels. In Proceedings of the ACM SIGKDD Conference on Knowledge Discovery and Data Mining, Beijing, China, August 2012.
  Bibtex source | Abstract | Download in pdf format | Code



• B. Hariharan, S. V. N. Vishwanathan and M. Varma. Efficient max-margin multi-label classification with applications to zero-shot learning. Machine Learning Journal, 88(1):127--155, 2012.
  Bibtex source | Abstract | Download in pdf format | Code



• V. Jain and M. Varma. Learning to re-rank: Query-dependent image re-ranking using click data. In Proceedings of the International World Wide Web Conference, Hyderabad, India, March 2011.
  Bibtex source | Abstract | Download in pdf format | Slides



• S. V. N. Vishwanathan, Z. Sun, N. Theera-Ampornpunt and M. Varma. Multiple kernel learning and the SMO algorithm. In Advances in Neural Information Processing Systems, Vancouver, B. C., Canada, December 2010.
  Bibtex source | Abstract | Download in pdf format | Code | Spotlight



• B. Hariharan, L. Zelnik-Manor, S. V. N. Vishwanathan and M. Varma. Large scale max-margin multi-label classification with priors. In Proceedings of the International Conference on Machine Learning, Haifa, Israel, June 2010.
  Bibtex source | Abstract | Download in pdf format | Code



• M. Varma and A. Zisserman. A statistical approach to material classification using image patch exemplars. IEEE Transactions on Pattern Analysis and Machine Intelligence , 31(11):2032--2047, November 2009.
  Bibtex source | Abstract | Download in pdf format



• A. Vedaldi, V. Gulshan, M. Varma and A. Zisserman. Multiple kernels for object detection. In Proceedings of the International Conference on Computer Vision, Kyoto, Japan, September 2009.
  Bibtex source | Abstract | Download in pdf format | Code



• M. Varma and B. R. Babu. More generality in efficient multiple kernel learning. In Proceedings of the International Conference on Machine Learning, Montreal, Canada, pages 1065--1072, June 2009.
  Bibtex source | Abstract | Download in pdf format | Code



• T. E. de Campos, B. R. Babu and M. Varma. Character recognition in natural images. In Proceedings of the International Conference on Computer Vision Theory and Applications, Lisbon, Portugal, February 2009.
  Bibtex source | Abstract | Download in pdf format | Download English and Kannada datasets



• M. Varma and D. Ray. Learning the discriminative power-invariance trade-off. In Proceedings of the IEEE International Conference on Computer Vision, Rio de Janeiro, Brazil, October 2007.
  Bibtex source | Abstract | Download in pdf format | www | Code | Errata
  Please see the errata regarding the Caltech experiments



• M. Varma and R. Garg. Locally invariant fractal features for statistical texture classification. In Proceedings of the IEEE International Conference on Computer Vision, Rio de Janeiro, Brazil, October 2007.
  Bibtex source | Abstract | Download in pdf format



• N. Adabala and M. Varma and K. Toyama. Computer aided generation of stylized maps. Computer Animation and Virtual Worlds, 18(2):133--140, May 2007.
  Bibtex source | Abstract | Download in pdf format



• M. Varma and A. Zisserman. A statistical approach to texture classification from single images. International Journal of Computer Vision: Special Issue on Texture Analysis and Synthesis, 62(1--2):61--81, April 2005.
  Bibtex source | Abstract | Download in pdf format



• M. Varma and A. Zisserman. Unifying statistical texture classification frameworks. Image and Vision Computing, 22(14):1175--1183, December 2004.
  Bibtex source | Abstract | Download in pdf format



• M. Varma. Statistical Approaches To Texture Classification. DPhil Thesis, University of Oxford, October 2004.
  Bibtex source | Abstract | Download in pdf format (~18 Mb)



• M. Varma and A. Zisserman. Estimating illumination direction from textublue images. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Washington, DC, volume 1, pages 179--186, June 2004.
  Bibtex source | Abstract | Download in pdf format



• M. Varma and A. Zisserman. Texture classification: Are filter banks necessary? In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Madison, Wisconsin, volume 2, pages 691--698, June 2003.
  Bibtex source | Abstract | Download in pdf format



• M. Varma and A. Zisserman. Statistical approaches to material classification. In Proceedings of the Indian Conference on Computer Vision, Graphics and Image Processing, Ahmedabad, India, pages 167--172, December 2002.
  Bibtex source | Abstract | Download in pdf format



• M. Varma and A. Zisserman. Classifying materials from images: to cluster or not to cluster? In Proceedings of the 2nd International Workshop on Texture Analysis and Synthesis, Copenhagen, Denmark, pages 139--144, June 2002.
  Bibtex source | Abstract



• M. Varma and A. Zisserman. Classifying images of materials: Achieving viewpoint and illumination independence. In Proceedings of the 7th European Conference on Computer Vision, Copenhagen, Denmark, volume 3, pages 255--271. Springer-Verlag, May 2002.
  Bibtex source | Abstract | Download in pdf format



• M. Varma and V. S. Varma. Computer simulation of evolution. In Proceedings of the GIREP-ICPE International Conference, Ljubljana, Slovenia, pages 138--150, August 1996.
  Bibtex source