|
Jaidev Gill I'm a recent graduate of the Engineering Physics program at the University of British Columbia in Vancouver. During my time at UBC I worked on understanding the learned representations of neural networks with Professor Christos Thrampoulidis. Recently, I have embarked on a new journey joining the department of Electrical Engineering and Computer Science at the University of Michigan to start my PhD in Electrical and Computer Engineering working with Professor Lisa Li. |
|
Education
PhD in Electrical and Computer Engineering, University of Michigan, Ann Arbor, MI, 2024-Present |
ResearchMy research interests include control, optimization, machine learning, signal processing, and information theory. |
|
Deep Learning-Based Correction and Unmixing of Hyperspectral Images for Brain Tumor Surgery
David Black*, Jaidev Gill*, Andrew Xie*, Benoit Liquet, Antonio Di leva, Walter Stummer, Eric Suero Molina, * Co-first authors. arXiv Design and implementation of a fully deep learning-based algorithm to correct and quantify PpIX in cancerous brain tissue. |
|
Engineering the Neural Collapse Geometry of Supervised-Contrastive Loss
Jaidev Gill, Vala Vakilian, Christos Thrampoulidis. ICASSP 2024 / AAAI SA 2024 / arXiv Prototypical representations of feature embeddings can force the representations of SCL to a desired geometry. |
|
Symmetric Neural-Collapse Representations with Supervised Contrastive Loss: The Impact of ReLU and Batching
Ganesh Ramachandra Kini, Vala Vakilian, Tina Behnia, Jaidev Gill, Christos Thrampoulidis. ICLR 2024 / ICML HiLD 2023 / DeepMath 2023 / arXiv In the presence of ReLU activations SCL learns representations robust to imbalances in the data distribution. |
Teaching
Teaching Assistant, PHYS 159 - Introductory Physics Laboratory for Engineers, UBC, Spring 2024
|
|
This website's source code is courtesy of John Barron. |