About Me
You can call me Prud. I am a PhD student in Computer Science at UC San Diego.
My research is on how neural networks (NN) acquire capabilities. I found neural networks learn tasks before the loss goes down. Currently, I am investigating circuits responsible.
Broadly I spend my days thinking whether NN's will gain meta-cognitive abilities as we scale thus explain their decisions.
More broadly I love reading books in botanical gardens & wondering what is truth and how do we justify it. Currently reading "Crescent moon and other stories" by Lao She (Translated in English).
If you have any insights into any of these, feel free to reach out.
Selected Works
Quiet Feature Learning in Algorithmic Tasks
Prudhviraj Naidu, Zixian Wang, Leon Bergen, Ramamohan Paturi
Pre-print
Scientific Document Retrieval using Multi-level Aspect-based Queries
Jianyou Wang, Kaicheng Wang, Xiaoyue Wang, Prudhviraj Naidu, Leon Bergen, Ramamohan Paturi
NeurIPS 2023
Metamorphic Testing for Convolutional Neural Networks: Relations over Image Classification
Prudhviraj Naidu, Hemanth Gudaparthi, and Nan Niu
2021 IEEE 22nd International Conference on Information Reuse and Integration for Data Science (IRI), 2021
