Aounon Kumar

I am a PhD student in the Computer Science department at the University of Maryland, working with Prof. Tom Goldstein and Prof. Soheil Feizi. My research is about making machine learning models certifiably robust against adversarial attacks. It involves devising methods that can generate provable guarantees for the predictions of a deep neural network in the presence of malicious corruptions of the input data that are unnoticeable to humans.

I have also worked on Approximation Algorithms and Approximation Hardness for Combinatorial Optimization problems. Before joining UMD, I completed my undergraduate studies at IIT Mandi and my master's at IIT Delhi.


Publications

Curse of Dimensionality on Randomized Smoothing for Certifiable Robustness
Aounon Kumar, Alexander Levine, Tom Goldstein, Soheil Feizi
ICML 2020, arXiv.

Certifying Confidence via Randomized Smoothing
Aounon Kumar, Alexander Levine, Soheil Feizi, Tom Goldstein
NeurIPS 2020, arXiv.

Detection as Regression: Certified Object Detection by Median Smoothing
Ping-yeh Chiang, Michael J. Curry, Ahmed Abdelkader, Aounon Kumar, John Dickerson, Tom Goldstein
NeurIPS 2020, arXiv.

Tight Second-Order Certificates for Randomized Smoothing
Alexander Levine, Aounon Kumar, Thomas Goldstein, Soheil Feizi
Preprint 2020, arXiv.

On the cost of essentially fair clusterings
Ioana O. Bercea, Martin Groß, Samir Khuller, Aounon Kumar, Clemens Rösner, Daniel R. Schmidt, Melanie Schmidt
APPROX 2019, arXiv.

Capacitated k-Center Problem with Vertex Weights
Aounon Kumar
FSTTCS 2016.

Recent

ICML 2020
NeurIPS 2020
NeurIPS 2020
Preprint 2020