Hi!
Hi!
Embedded Files
I am a fifth-year PhD candidate at University of California Santa Barbara. My research spans robust and fair machine learning. Specifically, I have been working on making DNN architectures robust to out-of-distribution corruptions and adversarial perturbations, focusing on image datasets. In parallel, I have also been looking into promoting long-term fairness in selection problems such as hiring or admissions. I am fortunate to be advised by Prof. Upamanyu Madhow and Prof. Ramtin Pedarsani.
I am generally interested in applying deep learning to problems in the space of vision and language models. In particular, I am intrigued by multimodal representation learning, self-supervision and generative models.
My background is diverse, with a demonstrated history of picking up challenging problems in different domains. I have pursued research internships at four different organizations (Meta, Amazon, Qualcomm, Nio), working on problems such as unsupervised personalization in multimodal encoders, tabular data synthesis using VAE and diffusion models, mapping for autonomous driving and deep learning approaches for in-cabin human presence detection.
I obtained a Master's degree from the Indian Institute of Science, Bangalore, with research focus on information and coding theory. I worked at MaxLinear Inc. for two years, developing signal processing algorithms for communication applications, after which I joined UC Santa Barbara in the fall of 2019 and transitioned to the machine learning space.
Contact: bpuranik[at]ucsb[dot]edu
News
News
[July 2024] Our work on long-term fairness in multi-agent selection problem has been accepted in the special issue on reliable and trustworthy ML at IEEE JSAIT. Here is a link to the paper on arxiv.
[Mar 2024] I will be spending the summer of 2024 in NYC as an applied scientist intern at Amazon AWS security analysis and AI research team!
[Jan 2024] Excited to announce our work on a new layer and approach for robustness in DNNs has been accepted to AISTATS 2024! Here is a link to the paper.
[Dec 2023] I will be spending the Spring of 2024 at Meta Reality Labs as a research scientist intern!
[Nov 2023] Our preprint on improving the robustness of DNNs through a communication-theory motivated approach is on arxiv.
[Jul 2023] Thrilled to be at ICML this year, presenting our paper on "Robust Deep Learning via Layerwise Tilted Exponentials" at the AdvML workshop! Thanks to my wonderful collaborators Ahmad, Yao and Madhow.
[Jun 2023] I will be spending this summer as a PhD Machine Learning intern at Nio USA, working on deep learning approaches for in-cabin localization, and exploring text-driven methods for image generation.
[May 2023] Excited to present our work on long-term fairness in sequential decision making in the Information-Theoretic Methods for Trustworthy Machine Learning Workshop at Simon's Institute, Berkeley.
[May 2023] Recipient of the Outstanding Teaching Assistant Award yet again from the ECE department
[Mar 2023] Successfully completed my PhD Qualification exam
[Jul 2022] Our work on "Generalized Likelihood Ratio Test for Adversarially Robust Hypothesis Testing" is accepted for publication at the IEEE Transactions on Signal Processing.
[Jun 2022] I will be presenting our work "Dynamic Positive Reinforcement for Long-Term Fairness" at the 2022 International Conference on Machine Learning (ICML) Workshop on Responsible Decision-Making in Dynamic Environments. Excited for my first in-person conference!
[Jun 2022] Thrilled to receive the Outstanding Teaching Assistant Award from the ECE department!
[Apr 2022] Our paper "A Dynamic Decision-Making Framework Promoting Long-Term Fairness" has been accepted for publication at the AAAI/ACM Conference on Artificial Intelligence, Ethics and Society (AIES), Oxford, 2022.
[Mar 2022] Our paper "Dynamic Positive Reinforcement for Long-Term Fairness" has been accepted to the ICLR 2022 Workshop on Socially Responsible Machine Learning.
[Dec 2021] Check out our preprint "Generalized Likelihood Ratio Test for Adversarially Robust Hypothesis Testing" on arXiv.
[Jun 2021] Looking forward to spending the summer of 2021 as an intern with Qualcomm's ADAS/Autonomy systems R&D team mentored by Urs Niesen & Meghana Bande. I will be working on approaches to build "maps" for autonomous driving.
[Jan 2021] Our paper "Adversarially Robust Classification based on GLRT" is accepted for publication at the IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) 2021.
Page updated
Google Sites
Report abuse