Rajeev Verma

I am a first year PhD student at AMLab / Delta Lab supervised by Eric Nalisnick. Previously, I studied Electrical Engineering at the Indian Institute of Technology Patna (IITP) and Artificial Intelligence at the University of Amsterdam (UvA).

My general research interests are in AI safety and responsible AI, and aim to build systems that are maximally useful to the society with provable guarantees. To this end, I am excited about uncertainty quantification and decision making, human-AI compatibility and complementarity, deferral systems, etc.

News

Jan 21, 2023 Learning to Defer to Multiple Experts: Consistent Surrogate Losses, Confidence Calibration, and Conformal Ensembles will be presented at AIStats’23. This paper is an extended version of the work presented earlier at the HMCaT workshop at ICML’22.
Jan 16, 2023 I started as a PhD candidate at AMLab / Delta Lab supervised by Eric Nalisnick.
Nov 1, 2022 New preprint: Learning to Defer to Multiple Experts: Consistent Surrogate Losses, Confidence Calibration, and Conformal Ensembles. We study the three Cs of deferral systems with multiple experts.
Sep 15, 2022 Talk at Innovation Center for Artificial Intelligence. Slides here.
Aug 25, 2022 Took a tortuous route, but I finally graduated from the University of Amsterdam with a master’s in Artificial Intelligence.

selected publications

  1. AISTATS
    Learning to Defer to Multiple Experts: Consistent Surrogate Losses, Confidence Calibration, and Conformal Ensembles
    Verma*, Rajeev, Barrejón*, Daniel, and Nalisnick, Eric
    In to appear in the proceedings of the 26th International Conference on Artificial Intelligence and Statistics (AISTATS) 2023
  2. ICML
    Calibrated Learning to Defer with One-vs-All Classifiers
    Verma, Rajeev, and Nalisnick, Eric
    In Proceedings of the 39th International Conference on Machine Learning (ICML) 2022