Robin Walters
I work on the role of symmetry in deep learning: how to build it into a model, when to relax it, and what it does to learning. I direct the Geometric Learning Lab in the Khoury College of Computer Sciences at Northeastern University.
The motivating problem is data efficiency. My lab builds models that use symmetry, geometry and compositional structure to extrapolate by design rather than by scale, which matters most in robotics, science and engineering, where data is expensive and a confident wrong answer is costly.
Before Northeastern I did a PhD in mathematics at the University of Chicago with Victor Ginzburg, and spent a year at Pixar working on rendering.
Updates
2026
Sep · Speaking at Mathematical Approaches to Challenges in Modern Artificial Intelligence, National University of Singapore, and at the Wallenberg Advanced Scientific Forum on symmetries in neural networks, Rånäs Castle, Sweden.
Congratulations to Neel Sortur and Purvik Patel, who have finished their master's degrees. Neel has joined AIC and Purvik has joined Braid Robotics.
Congratulations to Dian Wang, who has started as an Assistant Professor at the University of Macau, where he directs the Embodied Intelligence lab.
Received the Khoury College Junior Researcher Award, given to one pre-tenure faculty member each year.
Serving as Program Chair for Learning on Graphs (LoG) 2026.
Co-organising TAG-DS × Boston Symmetry Day 2026. Details and registration.
Jul · Talk at the SIAM Annual Meeting, special session on algebro-geometric approaches to deep learning, Cleveland.
Four papers at ICLR 2026, two at ICML 2026, one at RSS 2026. See the list.
Renaissance Philanthropy award for symmetry-aware generative AI for differential equations and mathematical discovery, with collaborators at the CfA, MIT and Harvard.
NSF CAIG award for HydrotherML, machine-learning-accelerated hydrothermal modelling for earth science, with Noah Randolph-Flagg (Kentucky).
RA-L Honorable Mention, five of 1,700 papers, for our work on language-conditioned manipulation.
2025
Congratulations to Dian Wang, who completed his PhD and has joined Stanford as a postdoc, having been named a JPMorgan PhD Fellow.
Congratulations to Eddie Berman and Luisa Li, named honorable mentions for the CRA Outstanding Undergraduate Researcher Award. Eddie has started his PhD at Harvard as an NSF GRFP Fellow and Luisa hers at Johns Hopkins.
Awarded the NSF CAREER grant for Improving Data Efficiency in Deep Learning with Relaxed Symmetry Constraints, $600,000 over 2025–2030.
Dec · Three posters at NeurIPS 2025: Bridging Equivariant GNNs and Spherical CNNs, A Practical Guide for Incorporating Symmetry in Diffusion Policy, and 3D Equivariant Visuomotor Policy Learning via Spherical Projection.
Nov · Talks at ICERM on designing neural networks to understand algebraic and geometric structure, and at Woods Hole Oceanographic Institution on equivariant networks for dynamics and control.
Oct · Talk at the Georgia Tech AI4Science seminar.
Jun · Talk at the EquiVision workshop at CVPR, Nashville.
May · Talk at the GRASP seminar, University of Pennsylvania.
Jan · Talk at the AMS special session, Joint Mathematics Meetings, Seattle.
Yuanyuan Yang has started her PhD in the lab.