Akshay Rangamani

Assistant Professor

Department of Data Science
Ying Wu College of Computing
New Jersey Institute of Technology
akshay [dot] rangamani [at] njit [dot] edu

I am looking to hire one PhD student for the Fall 2025 cycle. Reach out to me if you are interested in the science of deep learning. Keep in mind that I cannot respond to every email, but I will read them all.

I am organizing a Deep Learning Reading Group in Spring 2025. Reach out to me if you would like to join!

NJIT UG/MS Students - fill out this form if you're interested in working with me.

Research Overview

Recent Topics: Identifying sparse and low rank structures in deep networks (like Neural Collapse.)

I am broadly interested in the science of deep learning. I use theoretical and empirical approaches to understand and explain phenomena in deep learning. I have also worked on problems in sparse and low rank signal processing, with applications to computer vision, biomedical signal processing, and online social networks. I also dabble in problems in computational neuroscience.

Bio

I am an Assistant Professor in the Data Science department at NJIT. Between 2020 and 2023 I was a postdoc at the Center for Brains Minds and Machines at MIT, working with Prof. Tomaso Poggio. I got my PhD in Electrical and Computer Engineering at Johns Hopkins University, where I was a member of the Digital Signal Processing Laboratory, advised by Prof. Trac D. Tran. Before coming to JHU, I graduated with a B.Tech in Electrical Engineering from IIT Madras

Current Students

Altay Ünal (Data Science PhD Student)
Shweta Shardul (Computer Science MS Student)
Rohan Shanbag (Honors College BS Student)
Lakshya Chauhan (High School Student)

Teaching

DS677 - Deep Learning ('25 - Spring; '24 - Spring, Fall)

Selected Publications and Preprints

Year	Publication
2025	Low Rank and Sparse Fourier Structure in Recurrent Networks Trained on Modular Addition ICASSP 2025
2025	On Generalization Bounds for Neural Networks with Low Rank Layers ALT 2025 with Andrea Pinto and Tomaso Poggio
2023	Feature Learning in Deep Classifiers through Intermediate Neural Collapse ICML 2023 with Marius Lindegaard, Tomer Galanti and Tomaso Poggio
2023	Dynamics in Deep Classifiers trained with the Square Loss: normalization, low rank, neural collapse and generalization bounds RESEARCH with Mengjia Xu, Tomer Galanti, Qianli Liao and Tomaso Poggio
2023	For Interpolating Kernel Machines, Minimizing the Norm of the ERM Solution Maximizes Stability [pdf] Analysis and Applications 20th Anniversary Special Issue with Lorenzo Rosasco and Tomaso Poggio
2022	Neural Collapse in Deep Homogeneous Classifiers and The Role of Weight Decay IEEE ICASSP 2022 with Andrzej Banburski-Fahey
2021	A Scale Invariant Flatness Measure for Deep Network Minima IEEE ICASSP 2021 with Nam H. Nguyen, Abhishek Kumar, Dzung Phan, Sang H. Chin, Trac D. Tran
2019	Target Tracking and Classification Using Compressive Sensing Camera for SWIR Videos Signal Image and Video Processing with Chiman Kwan, Bryan Chou, Jonathan Yang, Trac Tran, Jack Zhang, Ralph Etienne-Cummings
2018	Reconstruction-free Deep Convolutional Neural Networks for Partially Observed Images IEEE GlobalSIP 2018 with Arun Nair, Luoluo Liu, Sang H. Chin, Muyinatu A. Lediju Bell and Trac D. Tran
2018	ChieF : A Change Pattern based Interpretable Failure Analyzer IEEE Big Data 2018 with Dhaval Patel, Lam Nguyen, Shrey Srivastava, and Jayant Kalagnanam
2018	Sparse Coding and Autoencoders (arXiv) [pdf] IEEE ISIT 2018 with Anirbit Mukherjee, Amitabh Basu, Trac D. Tran, Sang H. Chin
2018	A Greedy Pursuit Algorithm for Separating Signals from Nonlinear Compressive Observations IEEE ICASSP 2018 with Dung Tran, Trac D. Tran, Sang H. Chin
2016	Predicting Local Field Potentials with Recurrent Neural Networks IEEE EMBC 2016 with Louis Kim, Jacob Harer, Sang H. Chin, et. al.
2015	Targeted Dot Product Representation for Friend Recommendation in Online Social Networks IEEE/ACM ASONAM 2015 with Minh Dao, Nam P. Nguyen, Trac D. Tran, Sang H. Chin