Niru Maheswaranathan

Research Scientist, Reality Labs, Facebook

Joined

November 2017

Names

Niru Maheswaranathan (Preferred)

Emails

****@stanford.edu (Confirmed)

****@google.com (Confirmed)

****@hey.com (Confirmed)

****@fb.com (Confirmed)

****@niru.dev (Confirmed)

Personal Links

Homepage

Google Scholar

DBLP

Career & Education History

Research Scientist

Reality Labs, Facebook (meta.com)

2021 – Present

Research Engineer

Google Brain, Google (google.com)

2017 – 2021

PhD student

Neurosciences PhD Program, Stanford University (stanford.edu)

2011 – 2017

Undergrad student

Biomedical Engineering, Duke University (duke.edu)

2007 – 2011

Advisors, Relations & Conflicts

Coauthor

David Sussillo

2019 – Present

Coauthor

Alex H Williams

2016 – 2021

Coauthor

Jascha Sohl-Dickstein

2013 – 2021

Coauthor

Lane McIntosh

2013 – 2021

PhD Advisor

Surya Ganguli

2011 – 2017

PhD Advisor

Stephen Baccus

2011 – 2017

Expertise

deep learning

optimization

neural networks

Present

computational neuroscience

retina

Present

recurrent neural networks

Present

interpretability

Present

Publications

A mechanistically interpretable model of the retinal neural code for natural scenes with multiscale adaptive dynamics
Xuehao Ding, Dongsoo Lee, Satchel Grant, Heike Stein, Lane McIntosh, Niru Maheswaranathan, Stephen Baccus
- 14 May 2023 (modified: 14 May 2023)
- OpenReview Archive Direct Upload
- Readers: Everyone
Understanding How Encoder-Decoder Architectures Attend
Kyle Aitken, Vinay Venkatesh Ramasesh, Yuan Cao, Niru Maheswaranathan
- Published: 09 Nov 2021, Last Modified: 05 May 2023
- NeurIPS 2021 Poster
- Readers: Everyone
Reverse engineering learned optimizers reveals known and novel mechanisms
Niru Maheswaranathan, David Sussillo, Luke Metz, Ruoxi Sun, Jascha Sohl-Dickstein
- Published: 09 Nov 2021, Last Modified: 05 May 2023
- NeurIPS 2021 Poster
- Readers: Everyone
TaskSet: A Dataset of Optimization Tasks
Luke Metz, Niru Maheswaranathan, Ruoxi Sun, C. Daniel Freeman, Ben Poole, Jascha Sohl-Dickstein
- 28 Sept 2020 (modified: 05 May 2023)
- ICLR 2021 Conference Blind Submission
- Readers: Everyone
Overcoming barriers to the training of effective learned optimizers
Luke Metz, Niru Maheswaranathan, C. Daniel Freeman, Ben Poole, Jascha Sohl-Dickstein
- 28 Sept 2020 (modified: 05 May 2023)
- ICLR 2021 Conference Blind Submission
- Readers: Everyone
The geometry of integration in text classification RNNs
Kyle Aitken, Vinay Venkatesh Ramasesh, Ankush Garg, Yuan Cao, David Sussillo, Niru Maheswaranathan
- Published: 12 Jan 2021, Last Modified: 12 Oct 2025
- ICLR 2021 Poster
- Readers: Everyone
Reverse engineering learned optimizers reveals known and novel mechanisms
Niru Maheswaranathan, David Sussillo, Luke Metz, Ruoxi Sun, Jascha Sohl-Dickstein
- 28 Sept 2020 (modified: 05 May 2023)
- ICLR 2021 Conference Blind Submission
- Readers: Everyone
Guided evolutionary strategies: augmenting random search with surrogate gradients
Niru Maheswaranathan, Luke Metz, George Tucker, Dami Choi, Jascha Sohl-Dickstein
- 2019 (modified: 11 Nov 2022)
- ICML 2019
- Readers: Everyone
Understanding and correcting pathologies in the training of learned optimizers
Luke Metz, Niru Maheswaranathan, Jeremy Nixon, C. Daniel Freeman, Jascha Sohl-Dickstein
- 2019 (modified: 11 Nov 2022)
- ICML 2019
- Readers: Everyone
Recurrent Segmentation for Variable Computational Budgets
Lane McIntosh, Niru Maheswaranathan, David Sussillo, Jonathon Shlens
- 2018 (modified: 10 Nov 2022)
- CVPR Workshops 2018
- Readers: Everyone

View all 13 publications

Co-Authors

View all 28 co-authors