Aarshvi Gajjarmy photo


I am a third year Ph.D. student in the Algorithms and Foundations Group at NYU. My advisors are Christopher Musco and Chinmay Hegde.

I completed MS from UMass with Cameron Musco as my advisor. Prior to that, I worked as a Strat at Goldman Sachs and obtained my undergraduate degree from IIIT Hyderabad.


Research

My research centers around algorithms for data-limited problems. Specifically, questions that interest me are:

  • How many samples are required to approximately solve nonlinear regression?
  • How does this requirement vary for different nonlinear function classes?
I am also currently exploring how adding some noise to the Hessian impacts convergence of optimization algorithms.
In my research, I employ tools from theoretical computer science, high-dimensional statistics, and approximation theory.

Publications

All author names are in alphabetical order, unless denoted by †.

† Improved Bounds for Agnostic Active Learning of Single Index Models
Aarshvi Gajjar , Xingyu Xu, Chinmay Hegde and Christopher Musco
Under review, short version at RealML Workshop NeurIPS, 2023

Active Learning for Single Neuron Models with Lipschitz Non-Linearities
Aarshvi Gajjar, Chinmay Hegde and Christopher Musco
AISTATS, 2023, short version Spotlight at DLDE Workshop, NeurIPS 2022

Subspace Embeddings under Nonlinear Transformations
Aarshvi Gajjar, Cameron Musco
ALT (Conference on Algorithmic Learning Theory), 2021

Contact. [firstname] 'at' nyu 'dot' edu