Sumit Basu

"I am now a researcher at Microsoft Research in Redmond, WA working in the Knowledge Tools group, led by John Platt. Until September 2002, I was a graduate student in the Electrical Engineering and Computer Science Department at MIT, working with Professor Alex 'Sandy' Pentland at the Media Lab. My research is focused on developing machine learning methods for understanding patterns in complex data, in domains from human conversations to source code. I also am involved with the University of Washington, and co-teach the Markovia seminar on machine learning (cse590mv) there with Tanzeem Choudhury.

You can find more information about my current research at my Microsoft Research page, and see more about my music and past work by browsing the navigation topics to the left."

Work and research


Workshop on Data-Driven Education at NIPS 2013 along organized with Jonathan Huang and Kalyan Veeramachaneni. 

Here are the slides from the recent UW-MSR Colloquium Talk on "Making Reading More Effective" (November 2012), which gives an early overview of some of the work they are doing in the education space.

Presentation of thier NAACL 2012 paper on generating quizzes from arbitrary text data, part of their efforts on teaching with machine learning. The corpus from the paper as well as a video of the slideshow with narration will be available shortly.

Current projects

Teaching with Machine Learning: using machine learning to help students and teachers of all ages and all types of educational goals achieve their objectives more effectively and efficiently.

Sho: a powerful interactive environment for scientific computing and prototyping based on IronPython. Find out more and download it here. Also check out this code for getting real-time skeleton data from Kinect in Sho.

Earlier projects

Songsmith: a songwriting tool that takes melodies and helps develop accompaniments for them: based on this research with Dan Morris, it's now a product (with much help from the MSR Advanced Development Team). Check it out and download the trial here. It's also now free to many educational institutions via MSDN Academic Alliance and the Innovative Teachers' Network.

StickySorter: a tool for doing affinity diagramming and other flavors of information organization developed with Julie Guinn and Office Labs: you can download it here.

Music Analysis/Synthesis: using machine learning to help users understand, manipulate, and create music

Systems and Machine Learning: using machine learning to address problems in computer systems

Conversational Scene Analysis: seeking structure and content from conversational patterns

En savoir plusElectrical Engineering and Computer Science Department at MITMIT Media laboratory

  • Conversational Scene Analysis
  • Mania/Depression Assessment
  • The Influence Model
  • The Facilitator Room
  • Smart Headphones

The complete list and all information on these projects are here 


Recent papers

Find here the complete list of Sumit Basu's publications


Sumit Basu
One Microsoft Way
Redmond, WA 98052
Voice: 425.706.7971
Fax: 425.936.7329