Robin Walters

Robin Walters

Assistant Professor

Khoury College of Computer Sciences

Northeastern University

r.walters@northeastern.edu

Hi! I'm an Assistant Professor of computer science at Northeastern University and director of the Geometric Learning Lab. My research is on symmetry in deep learning.

Our research spans from theory of deep learning through the lens of symmetry to development of new equivariant models to applications of equivariant neural networks to a diverse set of applications in science and engineering.

Updates


---- 2025 ---

Awarded the NSF CAREER Grant for the project Improving Data Efficiency in Deep Learning with Relaxed Symmetry Constraints, receiving $600,000 for the period 2025-2030.

I will be speaking at the NeurReps Workshop in San Diego, CA in December 2-7, 2025.

I will be presenting Research on Equivariant Neural Networks for Dynamics and Control at the Woods Hole Oceanographic Research Institute in November 5, 2025.

Will speak about How to Design Neural Networks to Understand Algebraic and Geometric Structure at the upcoming ML Seminar held during The Institute for Computational and Experimental Research in Mathematics (ICERM) conference this November 4, 2025.

I will be presenting Research on Improving Convergence and Generalization in Deep Learning Using Parameter Symmetries at the GATech AI4Science Presentation in October 28, 2025.

Awarded the EAI Seed Funding of $60,000, in collaboration with Peter Schindler (MIE).

Awarded the Northeastern TIER 1 Interdisciplinary Seed Grant funding of $50,000, in collaboration with Peter Schindler (MIE).

I will present at the Equivision Workshop during CVPR in Nashville, TN in June 11, 2025.

I will give a talk at the GRASP Seminar at the University of Pennsylvania in April 22, 2025.

I will be presenting on the topic of Improving Convergence and Generalization in Deep Learning Using Parameter Symmetries during the Joint Mathematics Meetings AMS Special Session in Seattle, WA in January 18, 2025.

---- 2024 ---

I will present Research on Simulating Radar Using Equivariant Graph Neural Networks at the Recent Advances in AI for National Security Conference in Bedford, MA in December 16, 2024.

I will speak about pushing the limits of Equivariant Neural Networks at the IROS Workshop on Equivariant Robotics in Abu Dhabi, UAE in October 14, 2024.

I will share insights on pushing the limits of Equivariant Neural Networks at The AI Institute in Cambridge, MA in October 2024.

I will discuss pushing the limits of Equivariant Neural Networks during the NeurReps Global Speaker Series at MIT in October 23, 2024.

Recognized as an Outstanding Paper Finalist at the Conference on Robot Learning (CoRL), held November 6 to 9, 2024.

Awarded Best Paper at the 2nd Workshop on High-dimensional Learning Dynamics at ICLR, 2024.

I will discuss Equivariant Neural Networks at the Computer Vision Reading Group at CalTech in June 18, 2024.

I will discuss Equivariant Neural Networks at the Symposium on Graphics Processing Graduate School at MIT in June 22, 2024.

I will present Research at the AstroAI Seminar at the Harvard Smithsonian Center for Astrophysics in June 17-21, 2024.

I will deliver an oral session on Improving Convergence and Generalization in Deep Learning Using Parameter Symmetries at ICLR in Vienna, Austria in May 7 to May 11, 2024.

Show older updates
I am giving a tutorial at Symposium on Geometry Processing 2024 graduate summer school on June 22, 2024.
Please submit to the Geometry-grounded Representation Learning and Generative Modeling Workshop we are organizing at ICML 2024
Congratulations to Circe Hsu who won a PEAK award to fund her research on ML4Math.
Thanks to Northeastern Research Development for Interdisciplinary TEIR 1 Seed funding.
Paper on using homomorphic structure for rapid calibration of EMG devices accepted to TMLR
We are giving a tutorial at AI in Action on Everyday Robotics
Thanks to EAI Institute> for Seed funding with Peter Schindler to develop generative models for novel catalysts.
Check out this news article on GLL Undergrad Neel Sortur.
Talking at Boston Dynamics AI Institute.
I'll be giving a talk at the Open Neighborhood Seminar in the Harvard Math Department on Feb 7 at 4:30pm
2 papers accepted to ICLR 2024:
(Oral) Improving Convergence and Generalization Using Parameter Symmetries
Fourier Transporter: Bi-Equivariant Robotic Manipulation in 3D
4 papers accepted to NeurIPS:
1. A General Theory of Correct, Incorrect, and Extrinsic Equivariance
2. Modeling Dynamics over Meshes with Gauge Equivariant Nonlinear Message Passing
3. Equivariant Single View Pose Prediction Via Induced and Restricted Representations
4. Topological Obstructions and How to Avoid Them
We had a very successful Boston Symmetry Day with over 100 attendees from 30 different institutions!
Our paper on modeling radar using equivariant neural networks was accepted to NeurReps proceedings tracks.
See our labs work in progress at NeurReps, UniReps, AI4Science, and Learning on Graphs workshop where we present 8 different extended abstracts.
I'll be giving a talk at AstroAI at the Harvard-Smithsonian CfA on Jan 29, at 12PM
Talk at GATech AI/ML in Physics 2023 .
Colloquium at Rutgers University Mechanical and Aerospace Engineering .
Talk at Widely Applied Mathematics Seminar at Harvard SEAS .
Presented our work at Simon's Foundation Mathematical and Scientific Foundations of Deep Learning Annual Meeting. Part of our Scale-MoDL NSF Grant.
Awarded NSF FRR grant with Robert Platt.
Awarded Line Grant to collaborate with Rajmonda Caceres at MITLL and study radar signal processing with equivariant NN.
Honored to serve as AC for Learning on Graphs (LOG) 2024.
Honored to serve as AC for ICLR 2024.
Talk on Simulating Radar with ENNs at Graph Ex Symposium.
Talk and Tutorial at AIFAI summer school and workshop.
Talk at Umass Amherst Math department.
Talk at Michigan state masters.
Paper accepted ICML - Generative adversarial symmetry discovery.
Paper accepted at CorL.
Master student Sumukh graduates with Khoury Research Award.
Undergrad student Neel wins PEAK Research Award.
More details can be found in my CV.