Subhojit Som – Home Page


I am currently at Microsoft Azure Machine Learning where as an Applied Scientist I am working on Machine Learning algorithms and applications. Previously I was at Bing working on entity related projects. I worked on the Windows entity search project and Bing natural language conversational search project. The conversational search enabled users to have a conversation with Bing asking natural language questions about entities. The search engine can remember the context as the conversation progresses and resolve user questions about an entity (i.e., about which entity this question refers to) when the conversation involves several entities.

I obtained MS and PhD from The Ohio State University under the supervision of Prof. Lee C. Potter. At Ohio State I also worked with Prof. Philip Schniter. I received BTech (Hons) from Indian Institute of Technology, Kharagpur and worked at Texas Instruments for couple of years before joining grad school. Following my graduation from Ohio State I was a Postdoctoral Fellow at the School of Electrical and Computer Engineering of Georgia Institute of Technology for a year and worked with Prof. Mark A. Clements. My graduate research focused on sparse inference and its application to imaging. Sparse inference algorithms we can estimate signals that are sparse from significantly smaller number of measurements. I worked on Electron Paramagnetic Resonance Imaging which can help monitoring efficacy of cancer treatment and wound healing. I proposed a new parametric imaging model and used sparse inference methods to achieve an order of magnitude reduction in data acquisition time. As a consequence the patient has to spend less amount of time in the scanner. I have also worked on fast inference algorithms for structured sparse signals where the signal is not just sparse but there also exists some pattern in that sparsity. Exploiting this pattern can help us to estimate the signal from even lesser number of measurements with same estimation accuracy.