The SSB prize is a coveted prize for multidisciplinary science and engineering in
SIGCHI societal impact award in 2017:
MacArthur grant in 2016:
Polya prize in 2014:
Knuth prize in 2011:
Computer Aided Verification award in 2011:
These, and the other honors our researchers have won over the years are indeed a source of great pride for us. At the same time, we recognize that almost none of this would have been possible without the great collaborations we have with our academic partners across the globe. MSR, since inception, has always recognized the importance of collaborations with academic partners and other researchers, not only for advancing the state-of-the-art, but also for their incisive feedback on our research quality and directions. In fact,
Let me use Varma's (our most recent honoree) work as an example. He got his PhD from
A man smiling at the cameraManik Varma, researcher in the
However, his most impressive and impactful of contribution over the past decade is in Extreme Classification- a sub-discipline of Machine Learning which Manik helped originate. Instead of traditional classification algorithms, which classify objects with a small number of labels (categories), Varma has been building classifiers which classify objects with a large space of labels- think millions of labels! This formulation with a large space of labels makes sense for several scenarios such as phrase selection (for advertiser bidding, for example) in online advertisements, and tagging Wikipedia articles with tags, where the space of phrases or tags is very large. The initial algorithms used an ensemble of trees with lots of tricks to help scale while maintaining accuracy. More recent algorithms are based on embeddings generated by Deep Neural Networks.
He started this area and, together with colleagues at Microsoft and students at IIT Delhi, wrote the first papers on this topic. Over the past decade, we have seen papers by Varma and his collaborators on this topic routinely appearing in conferences such as ICML, KDD, NIPS, SIGIR, WSDM and WWW, with some of the papers winning best paper awards and receiving considerable attention and recognition from the community. More than ten workshops have been organized in the last seven years in various conferences on this topic, and he has been the keynote speaker in most of these workshops.
He also collected benchmarks and created an 'Extreme Classification Repository', which has been viewed and downloaded thousands of times every month and is now used as a standard to measure algorithmic progress in this area. In addition, he has been a key contributor in building ML algorithms that run on very small devices, with very low memory and power footprint.
Varma is an ambassador for
We live in interesting times where industry and academia have a great opportunity to work together to make scientific progress. Some problems that arise in industrial practice are open ended and need many years of sustained research to make progress in solving them. Exposure to such problems greatly benefits students and faculty in academia and hence such collaborations are mutually beneficial.
MSR is committed to continuing collaborations with our academic colleagues, to expose and formulate difficult problems that arise in industrial settingsand bringing to bear the collective wisdom of the academic community and advancing the state of the art to solve such problems.
For more information on
(C) 2020 Electronic News Publishing, source