Akshay Rangamani

EmailAMR239njitedu

h-index

Research activity per year

Fingerprint

Dive into the research topics where Akshay Rangamani is active. These topic labels come from the works of this person. Together they form a unique fingerprint.

1 Similar Profiles

10 Conference contribution
6 Article
3 Conference article

Low Rank and Sparse Fourier Structure in Recurrent Networks Trained on Modular Addition
Rangamani, A., 2025, In: ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings.
Research output: Contribution to journal › Conference article › peer-review
On Generalization Bounds for Neural Networks with Low Rank Layers
Pinto, A., Rangamani, A. & Poggio, T., 2025, In: Proceedings of Machine Learning Research. 272, p. 921-936 16 p.
Research output: Contribution to journal › Conference article › peer-review
Dynamics in Deep Classifiers Trained with the Square Loss: Normalization, Low Rank, Neural Collapse, and Generalization Bounds
Xu, M., Rangamani, A., Liao, Q., Galanti, T. & Poggio, T., 2023, In: Research. 6, 0024.
Research output: Contribution to journal › Article › peer-review

Open Access
27 Link opens in a new tab Scopus citations
Feature Learning in Deep Classifiers through Intermediate Neural Collapse
Rangamani, A., Lindegaard, M., Galanti, T. & Poggio, T., 2023, In: Proceedings of Machine Learning Research. 202, p. 28729-28745 17 p.
Research output: Contribution to journal › Conference article › peer-review
35 Link opens in a new tab Scopus citations
For interpolating kernel machines, minimizing the norm of the ERM solution maximizes stability
Rangamani, A., Rosasco, L. & Poggio, T., Jan 1 2023, In: Analysis and Applications. 21, 1, p. 193-215 23 p.
Research output: Contribution to journal › Article › peer-review
5 Link opens in a new tab Scopus citations

New insights into training dynamics of deep classifiers
Rangamani, A.
7/4/23
1 item of Media coverage
Press/Media: Press / Media
Studies from Massachusetts Institute of Technology Describe New Findings in Science and Technology (Dynamics in Deep Classifiers Trained with the Square Loss: Normalization, Low Rank, Neural Collapse, and Generalization Bounds)
Xu, M. & Rangamani, A.
3/29/23
1 item of Media coverage
Press/Media: Press / Media
New insights into training dynamics of deep classifiers
Rangamani, A.
3/15/23
1 item of Media coverage
Press/Media: Press / Media
-MIT - New insights into training dynamics of deep classifiers
Rangamani, A.
3/9/23
2 items of Media coverage
Press/Media: Press / Media
New insights into training dynamics of deep classifiers
Xu, M. & Rangamani, A.
3/8/23 → 3/9/23
2 items of Media coverage
Press/Media: Press / Media

Akshay Rangamani

Fingerprint

Collaborations and top research areas from the last five years

Dive into details

Low Rank and Sparse Fourier Structure in Recurrent Networks Trained on Modular Addition

On Generalization Bounds for Neural Networks with Low Rank Layers

Dynamics in Deep Classifiers Trained with the Square Loss: Normalization, Low Rank, Neural Collapse, and Generalization Bounds

Feature Learning in Deep Classifiers through Intermediate Neural Collapse

For interpolating kernel machines, minimizing the norm of the ERM solution maximizes stability

New insights into training dynamics of deep classifiers

Studies from Massachusetts Institute of Technology Describe New Findings in Science and Technology (Dynamics in Deep Classifiers Trained with the Square Loss: Normalization, Low Rank, Neural Collapse, and Generalization Bounds)

New insights into training dynamics of deep classifiers

-MIT - New insights into training dynamics of deep classifiers

New insights into training dynamics of deep classifiers

Akshay Rangamani

Fingerprint

Collaborations and top research areas from the last five years

Research output

Press/Media