Location: E18-304

Abstract: Moravec’s Paradox observes that AI systems have struggled far more with learning physical action than symbolic reasoning. Yet just recently, there has been a tremendous increase in the capability of AI-driven robotic systems, reminiscent  of the early acceleration in language modeling capabilities a few years prior.  Using the lens of control-theoretic stability, this talk will demonstrate an exponential separation between natural regimes for learning in the physical world and in discrete/symbolic settings, thereby providing a mathematical basis for Moravec’s famous observation.  We then explain the recent progress in robot learning by  establishing that the innovations that immediately preceded these advances—prediction of open-loop “action-chunks”, and use of generative models, such as diffusion models, to parametrize the conditional distribution of robot actions—directly mitigate the aforementioned  difficulties. While our understanding of action chunking is rigorous, our findings regarding generative modeling are mainly empirical in nature. Thus, we conclude with open questions regarding how,  under what conditions, and by what mechanisms popular generative models enjoy the properties uncovered in our experimental study.  

Bio: Max Simchowitz is an assistant professor at the Machine Learning Department at Carnegie Mellon University with a courtesy appointment in the Robotics Institute. His work studies theoretical foundations and new methodologies for machine learning problems with an interactive, sequential, or dynamical component, currently focusing on reinforcement learning and applications to robotics. His past work has ranged broadly across control, theoretical reinforcement learning, optimization and algorithmic fairness. He received his PhD from University of California, Berkeley in 2021 under Ben Recht and Michael I. Jordan, and completed his postdoctoral research under Russ Tedrake in the Robot Locomotion Group at MIT. His work has been recognized with an ICML 2018 Best Paper Award, ICML 2022 Outstanding Paper Award, and RSS 2023 and ICRA 2024 Best Paper Finalist designations. 

Abstract: Spectral gradient methods, such as the recently proposed Muon optimizer, are a promising alternative to standard gradient descent for training deep neural networks and transformers. Yet, it remains unclear in which regimes these spectral methods are expected to perform better. In this talk, I will present a simple condition that predicts when a spectral update yields a larger decrease in the loss than a standard gradient step. Informally, this criterion holds when, on the one hand, the gradient of the loss with respect to each parameter block has a nearly uniform spectrum—measured by its nuclear-to-Frobenius ratio—while, on the other hand, the incoming activation matrix has low stable rank. It is this mismatch in the spectral behavior of the gradient and the propagated data that underlies the advantage of spectral updates. Reassuringly, this condition naturally arises in a variety of settings, including random feature models, neural networks, and transformer architectures. I will conclude by showing that these predictions align with empirical results in synthetic regression problems and in small-scale language model training.

Biosketch: Dmitriy Drusvyatskiy received his PhD from Cornell University in 2013, followed by a post doctoral appointment at University of Waterloo, 2013-2014. He joined the Mathematics department at University of Washington as an Assistant Professor in 2014 and was promoted to Full Professor in 2022. Since 2025, Dmitriy is a Professor at the Halıcıoğlu Data Science Institute (HDSI) at UC San Diego. Dmitriy’s research broadly focuses on designing and analyzing algorithms for large-scale optimization problems, primarily motivated by applications in data science. Dmitriy has received a number of awards, including the Air Force Office of Scientific Research (AFOSR) Young Investigator Program (YIP) Award, NSF CAREER, SIAG/OPT Best Paper Prize 2023, Paul Tseng Faculty fellowship 2022-2026, INFORMS Optimization Society Young Researcher Prize 2019, and finalist citations for the Tucker Prize 2015 and the Young Researcher Best Paper Prize at ICCOPT 2019.

TBD

TBD

TBD

TBD

TBD