Mohammed Ehsan Hoque

Graduate Student, MIT Media Laboratory

E15-446, 20 Ames Street, Cambridge, MA 02142-1308

Office Phone: (617) 324-2136

Email: mehoque AT {mit dot edu} or {gmail dot com}

• I was the finalist of the ASSETS08 Student Research Competition (SRC), and will compete at the SRC grand finals.

• I (along with Stefanie Shattuck-Hufnagel from the speech communications group) am organizing an informal weekly (Thursday afternoons from 5:15 pm to 6:15 pm) reading and discussion group to understand prosody. Read details here.
• My paper titled "Analysis of Speech Properties of Neurotypicals and Individuals Diagnosed with Autism and Down Syndrome" gets accepted at the ASSETS08, along with ACM travel grant award. [more updates]

Academia:

My research interests broadly revolve around speech processing, machine learning, emotion sensing, and computer vision.

I work as a Research Assistant at the Affective Computing Group of MIT Media Lab headed by Rosalind Picard. Our efforts currently involve development and evaluation of wearable social-emotional technology that helps individuals with high-functioning autism and/or Asperger syndrome to improve their social abilities. This work is a collaborative effort between the Massachusetts Institute of Technology Media Laboratory and the Groden Center.  

My masters thesis involved automatic recognition of intentions, i.e., dialogue acts, from conversational speech. For example, "Can you please pass me the salt?" does not really inquire whether the person is capable of passing the salt or not, but, instead, conveys an implicit request. In my thesis, I have developed and compared models to recognize 13 types of dialogue acts using prosody, discourse and with their fusion. Please read my thesis for further details.

During my undergrad at Penn State University, I was involved in building a Human Interactive Robot able to locate and recognize faces with some help from voice identification, as part of my senior design project. The system performed surprisingly well with an accuracy rate of 85% on a half-dozen pre-trained people and competed strongly against systems from other schools, including Carnegie Mellon among others at the 4^th Annual Student Design Contest organized by Rochester Institute of Technology, New York. Our project also won the "Best Design Award" at Penn State judged by the local industries. Details can be found here.

Industry:

I have worked for Goldman, Sachs & Co. as a Technology Analyst with the Global Control Technology Group under the Firmwide Technology division.

I was responsible for building a web-enabled account maintenance tool and was involved in all cycles of development of this tool. This tool freed up .25 head count per year in our group and currently being leveraged in the New York, London, Hong Kong, Tokyo and Bangalore offices.