About Me

Hello! I’m a fourth-year Computer Science PhD student at The University of Maryland (UMD), College Park. I’m very fortunate to be advised by Prof. Soheil Feizi, at the Center for Machine Learning.

My research focuses on improving the reliability, security, and robustness of Deep Neural Networks by exploring and characterizing their vulnerabilities and failure modes. I’m enthusiastic to improve our understanding of the over-sensitivity of modern Vision models and Large Language models to adversarial or jailbreaking inputs, as well as their apparent under-sensitivity to blind spots. Towards this, I have also worked on developing efficient attacks and robust training techniques that minimize computational overhead. More recently, I’ve been excited to investigate the phenomenon of hallucinations in Large Language Models and identify strategies for detecting them in real-world production settings.

Before joining UMD as a PhD student, I earned both my Bachelor’s and Master’s degrees in Mathematics at The Indian Institute of Science (IISc), Bangalore. I was fortunate to be advised by Prof. Venkatesh Babu at the Computational and Data Sciences Department, and by Prof. Kaushal Verma at the Mathematics Department during my studies at IISc.

Preprints & Publications

LLM-Check: Investigating Detection of Hallucinations in Large Language Models

G Sriramanan, S Bharti, VS Sadasivan, S Saha, P Kattakinda, S Feizi

Neural Information Processing Systems (NeurIPS), 2024

Fast Adversarial Attacks on Language Models In One GPU Minute

VS Sadasivan, S Saha*, G Sriramanan*, P Kattakinda, A Chegini, S Feizi

International Conference on Machine Learning (ICML), 2024

Exploring Geometry of Blind Spots in Vision Models

S Balasubramanian*, G Sriramanan*, VS Sadasivan, S Feizi

Neural Information Processing Systems (NeurIPS), 2023

Spotlight Paper

Toward Efficient Robust Training against Union of Lp Threat Models

G Sriramanan, M Gor, S Feizi

Neural Information Processing Systems (NeurIPS), 2022

Oral Paper at AdvML Frontiers Workshop @ ICML 2022

Scaling Adversarial Training to Large Perturbation Bounds

S Addepalli*, S Jain*, G Sriramanan, RV Babu

European Conference on Computer Vision (ECCV), 2022

Towards Efficient and Effective Adversarial Training

G Sriramanan*, S Addepalli*, A Baburaj, RV Babu

Neural Information Processing Systems (NeurIPS), 2021

Towards Achieving Adversarial Robustness Beyond Perceptual Limits

S Addepalli*, S Jain*, G Sriramanan, S Khare, RV Babu

Adversarial Machine Learning (AML) Workshop @ ICML 2021

Efficient Training Methods for Achieving Adversarial Robustness Against Sparse Attacks

S Addepalli*, D Behl*, G Sriramanan, RV Babu

Adversarial Robustness in the Real World (AROW) Workshop @ ECCV 2021

Boosting Adversarial Robustness Using Feature Level Stochastic Smoothing

S Addepalli*, S Jain*, G Sriramanan*, S Khare, RV Babu

Safe Artificial Intelligence for Automated Driving (SAIAD) Workshop @ CVPR 2021

Guided Adversarial Attack for Evaluating and Enhancing Adversarial Defenses

G Sriramanan*, S Addepalli*, A Baburaj, RV Babu

Neural Information Processing Systems (NeurIPS), 2020

Spotlight Paper

Towards achieving adversarial robustness by enforcing feature consistency across bit planes

S Addepalli*, BS Vivek*, A Baburaj, G Sriramanan, RV Babu

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020

Activities

Conference and Journal Reviews

I’ve served as a reviewer for ICCV 2021, CVPR 2022, ECCV 2022, ACML 2022, NeurIPS 2022/23, ICLR 2023, JMLR 2022/23 over the past few years.
I received the Outstanding Reviewer Award at CVPR 2022, ECCV 2022 and ICLR 2023.

Workshop Program Committees

I’ve served on the Program Committee of various Deep Learning workshops on Adversarial Robustness, Security and Socially Responsible ML over the past few years: ECCV 2020 2022, ICLR 2021 2022, CVPR 2021 2022, ICML 2021 2022, ICCV 2021, 2023.