Himanshu Singh

B611, R and D Block

IIIT Delhi

Okhla Phase 3, Delhi 110020

I am a researcher in the field of Machine Learning, currently pursuing my Ph.D. at IIIT Delhi where I am a part of Visual Conception Group working under the guidance of Dr. AV Subramanyam. With a specific focus on adversarial attacks, I delve deep into understanding the vulnerabilities of machine learning models and devising robust defenses against malicious manipulations.

Prior to my doctoral journey, I gained invaluable industry experience as a Research Scientist at Animaker. Working in a dynamic environment, I honed my skills in applying cutting-edge machine learning techniques to real-world problems. I worked on building the text to video tool Steve AI This experience provided me with a practical perspective and strengthened my ability to bridge the gap between academia and industry.

Thank you for visiting my website. I am always open to exciting opportunities and collaborations. Feel free to reach out to me if you have any questions or would like to discuss potential collaborations. I look forward to connecting with you!

selected publications

Language Guided Adversarial Purification

Himanshu Singh, and A V Subramanyam

In ICASSP 2024 - 2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Apr 2024

Abs IEEE arXiv Code

Adversarial purification using generative models demonstrates strong adversarial defense performance. These methods are classifier and attack-agnostic, making them versatile but often computationally intensive. Recent strides in diffusion and score networks have improved image generation and, by extension, adversarial purification. Another highly efficient class of adversarial defense methods known as adversarial training requires specific knowledge of attack vectors, forcing them to be trained extensively on adversarial examples. To overcome these limitations, we introduce a new framework, namely Language Guided Adversarial Purification (LGAP), utilizing pre-trained diffusion models and caption generators to defend against adversarial attacks. Given an input image, our method first generates a caption, which is then used to guide the adversarial purification process through a diffusion network. Our approach has been evaluated against strong adversarial attacks, proving its effectiveness in enhancing adversarial robustness. Our results indicate that LGAP outperforms most existing adversarial defense techniques without requiring specialized network training. This underscores the generalizability of models trained on large datasets, highlighting a promising direction for further research.