I'm a PhD student in Computer Science at the University of Wisconsin-Madison working on machine learning, specifically in the areas of efficient retrieval, value alignment, and pluralistic generative models. During my masters at the University of Washington, I worked on making learned representations and search indexing structures more flexible and efficient through approaches like Matryoshka Representation Learning, which allows a single embedding to adapt to different computational constraints. During my PhD, I've been exploring pluralistic approaches to aligning generative models to human values, developing frameworks that can capture diverse human preferences rather than assuming universal values.
I'm particularly interested in making AI systems that are both computationally efficient and aligned with human values. My work on adaptive representations has shown promising results for tasks like large-scale retrieval and few-shot learning, while maintaining robustness. On the alignment side, I'm working on models that can learn reward functions that reflect the plurality of human preferences while remaining efficient to train and deploy.
S. Velusamy, Rishubh Parihar, Raviprasad Kini, Aniket Rege
2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) 2020
P. Kharat, Aniket Rege, A. Goel, M. Kulkarni
International Conference on Cryptography, Security and Privacy 2018
Gantavya Bhatt, Aniket Rege, Aditya Sinha, Vivek Ramanujan, William Howard-Snyder, Kaifeng Chen, Prateek Jain, Ali Farhadi
S. Velusamy, Rishubh Parihar, Raviprasad Kini, Aniket Rege