anmol at mit dot edu
Research Projects
Social Sensing to Model the Evolution of Opinions
This is my PhD dissertation project. The goal is model the adoption of various social behaviors using mobile phone sensor datasets. For this project we have collected one year's worth of interaction data from an MIT dorm-- which includes over 65,000 call logs, 3.5 million proximity scans and 2.5 million location scans.
Recent papers discuss the spread of political opinions within this community. Currently upcoming analysis looks at the propagation of flu and social relationships. You can find more information here.
Predicting Social Outcomes from Vocal Tone & Prosody (past)
This was my Master's Thesis project at MIT. We used non-linguistic speech features to predict interest, engagement and objective outcomes of conversations, dates, tourist experiences and job interviews. More information about this project is available here.
The Jerk-o-Meter -- Real Time Speech feature Analysis (past)
A phone / PDA application that did real-time speech feature analysis. Special mention here because this got a ridiculous amount of press, and launched my Internet celebrity career. Shown here running on a Sharp Zaurus, photo credit Webb Chapell. More information available here.
Neil Gershenfeld's legendary course at the Media Lab-- I think this should be mandatory for all MLers. My class notes for reference available here, and more details about setting up your own FabLab here.
I worked on the first concept-car design iteration with the Smart Cities group. My contributions were around multihop networking and using e-ink as an electronic skin. This project has gone through many versions since then, and you see the latest incarnation here.