AstroAI Lunch Talks - January 29, 2024 - Robin Walters
29 Jan 2024 - Joshua Wing
The video can be found here: https://www.youtube.com/watch?v=HfKQBn0Yhq4
Speaker: Robin Walters (Northeastern University)
Title: Pushing the Limits of Equivariant Neural Networks
Abstract: Despite the success of deep learning, there remain challenges to progress. Deep models require vast datasets to train, can fail to generalize under surprisingly small changes in domain, and lack guarantees on performance. Incorporating symmetry constraints into neural networks has resulted in models called equivariant neural networks which have helped address these challenges. I.will discuss several applications, such as vehicle trajectory prediction, fluid dynamics modeling, radar signal processing, computer vision, and robotic grasping and pick and place tasks. However, there are also limits to the effectiveness of current ENNs. In many applications symmetry is only approximate, does apply across the entire input distribution, or cannot be easily encoded into the model. I will discuss solutions to these problems such as relaxed equivariance and models which take advantage of latent symmetry.
Bio: Robin Walters is an assistant professor in the Khoury College of Computer Sciences at Northeastern University, where he leads the Geometric Learning Lab. Robin’s research seeks to develop a fundamental understanding of the role symmetry plays in deep learning and to exploit this to improve the generalization and data efficiency of deep learning methods. This includes designing equivariant neural networks, symmetry discovery methods, and creating a theory of symmetry for model parameters. He has applied these methods to improve models in domains with complex dynamics including climate science, transportation, and robotics.
Watch the talk below!