Debajyoti Datta (Debo)


When the butterfly refused to fly away. At the American museum of Natural History, NY.

I am a graduate student at the University of Virginia, where I work in Differential Geometry and Tensor Decomposition.

High dimensional data often exhibit interesting geometric properties, and understanding the geometry can provide us insights into domain-specific applications (in education and healthcare). I work with the school of Psychology (to understand depression and anxiety), the School of Medicine, and the Curry School of Education. I am currently the co-chair of the BigScience Biomedical group and a contributor to the prompt engineering and bias-fairness group.

After my MS, I worked at on many language understanding tasks like intent classification and named entity recognition. Then for a year, I was at Crossing Minds working on recommendation systems (mostly on large scale matrix factorization). Now back at grad school working with the awesome people at the University of Virginia at Sensing Systems for Health Lab.


Some recent news (2021)

  • Dec : Will present our work on weak-supervision for data collection and expert in the loop work at the MathAI4Ed @NeurIPS 2021 and HCAI @NeurIPS 2021.
  • Dec : Upcoming talk at MIT Teaching Systems Lab.
  • Nov : Guest Lecturer in Mobile Sensing Class. Will talk about prompt engineering.
  • Nov : AERA 1 session and 2 papers accepted to AERA!
  • Oct : TADA Going to present our work on language examples at the decision boundary
  • Sept : AllenAI Hackathon We won the AllenAI Hackathon! Our project used Weak-Supervision to identify errors in the Toxicity Dataset.
  • July : LOGML SUMMER SCHOOL Attended the LOGML summer school. In our project we showed how probabilistic methods generated better geodesic interpolations than non-probabilistic methods. Implementation of this paper.
  • April: Will present our work at the EACL Workshop