CPS Events
Memristors for a Brain-Scale Neuromorphic Chip
Abstract
Recently, applications of artificial intelligence (AI) have far outpaced Moore’s law in chip development, thus creating an increasingly large gap between user demand and the supply that the semiconductor industry can deliver. In this talk, we will discuss the unique roles of memristor technologies that can be leveraged to develop scaled-up AI neural networks, particularly spiking neural networks (SNNs) for brain-like neuromorphic computing and unsupervised learning with high energy efficiency. Open-source memristor circuit designs, along with open-source software, may facilitate the development of micro- and nano-electronic systems that emulate brain functions. In this venue, we will discuss how to harness memristor-based circuits and systems to build memristor neurons, synapses, and their interconnects for ultra-high packing density, low power consumption, and the fabrication services needed to enable innovation.
Speaker Bio
Sung-Mo “Steve” Kang is a Distinguished Professor Emeritus and Research Professor at the Baskin School of Engineering, UC Santa Cruz; Chancellor Emeritus of UC Merced; and President Emeritus of KAIST. He has published more than 500 journal and conference papers, authored 10 books, and holds 17 patents. Before returning to academia in 1985, he led the development of the world’s premier fully CMOS 32-bit VLSI microprocessor chipsets for telecommunications and computing applications as a technical supervisor at AT&T Bell Laboratories in Murray Hill, New Jersey. This work was recognized as an IEEE Milestone in February 2025. He has received honors, including best paper awards, induction into the Silicon Valley Engineering Hall of Fame, the Alexander von Humboldt Senior US Scientists Award, the IEEE Millennium Medal, the IEEE Mac Van Valkenburg Circuits and Systems (CAS) Society Award, the IEEE CAS Society Technical Excellence Award, the US Semiconductor Research Corporation (SRC) Technical Excellence Award, the IEEE Leon K. Kirchmayer Graduate Teaching Technical Field Award, and the IEEE CAS Society John Choma Education Award, as well as the Chang-Lin Tien Education Leadership Award. Dr. Kang is a Life Fellow of the IEEE and a Fellow of the Association for Computing Machinery (ACM), the American Association for the Advancement of Science (AAAS), and the Asia-Pacific AI Association. He is a life member of the European Academy of Sciences and Arts and the Korean Academy of Science and Technology, and a foreign member of the National Academy of Engineering, Korea. He received his B.S. from Fairleigh Dickinson University, Teaneck, New Jersey, in 1970; an honorary B.S. from Yonsei University; an M.S. from the State University of New York at Buffalo in 1972; and a Ph.D. from the University of California at Berkeley in 1975, all in electrical engineering.
Neurosymbolic Approaches for Trustworthy, Explainable Complex Systems
Abstract
Complex systems are prone to errors and failures without knowing why. In critical domains like driving, these autonomous counterparts must be able to recount their actions for safety, liability, and trust. In this talk, I will argue that an explanation: a model-dependent reason or justification for the decision of the autonomous agent being assessed, is a key component for post-mortem failure analysis and pre-deployment verification. I will present two neurosymbolic frameworks that use a model and common sense knowledge to detect and explain unreasonable and unknown scenarios, even if they have not seen that error before. In the second part of the talk, I will motivate the use of neurosymbolic systems in application domains, including Large Language Models (LLMs). I will conclude by discussing new challenges in developing trustworthy, neurosymbolic AI towards complex systems that are explainable by design.
Speaker Bio
Leilani H. Gilpin is an Assistant Professor in the Department of Computer Science and Engineering at UC Santa Cruz. Her research focuses on the design and analysis of methods for autonomous systems to explain themselves. Her work has applications to robust decision-making, system debugging, and accountability. She holds a PhD in Computer Science from MIT, an M.S. in Computational and Mathematical Engineering from Stanford University, and a B.S. in Mathematics (with honors), B.S. in Computer Science (with highest honors), and a music minor from UC San Diego. Outside of research, Leilani enjoys swimming, cooking, hiking, emacs, and org mode.
Rare but Ruinous: The Geometry of Reliability for Complex Cyber-Physical Systems
Abstract
Reliable operation of cyber-physical systems from power grids and engineered infrastructure to autonomous-vehicle fleets hinges on our ability to estimate the probability of rare, high-consequence failures in high-dimensional state spaces. Classical tools struggle in this regime: Monte Carlo is statistically inefficient, scenario approximation delivers conservative certificates without rigorous error control, and many well-known analytical bounds rest on log-concavity assumptions that modern stochastic systems often violate. This talk offers a unifying geometric perspective: every rare-event estimator can be viewed as a path between two probability measures on a statistical manifold, and by Chentsov's theorem, this path has a canonical length.
The talk is divided into three parts. First, we show that optimal importance-sampling proposals can be understood as Fisher–Rao geodesics on the underlying statistical manifold. Second, we argue that several widely used safe approximations of chance constraints can be recovered as first-order Taylor expansions of those geodesics, and that these methods were, in effect, implicitly climbing this geometry all along. Third, we outline open problems for cyber-physical systems whose state space mixes continuous and discrete components, where the manifold is layered, and the geodesic can jump between layers.
Speaker Bio
Yury Maximov is Chief Science and Special Projects Officer at ZeroAvia, the company developing the first practical hydrogen-electric powertrains for commercial aviation. Previously, Yury was a Staff Scientist in the Theoretical Division at Los Alamos National Laboratory (LANL), working on optimization, machine learning, and stochastic methods for power grid reliability and renewable energy integration. Earlier, he held postdoctoral positions at Université Grenoble-Alpes and INRIA (France). Yury holds a Ph.D. in Applied Mathematics and Control. His research sits at the intersection of complex energy systems, optimization, and control.
Dynamical Signatures: Harnessing the Hidden Language of In-Space Electric Propulsion
Abstract
Low-thrust space electric propulsion systems offer long propulsion system lifetimes for satellite maintenance maneuvers. These thrusters operate by generating and accelerating plasmas, making the thrusters throttleable, propellant-efficient, and scalable from low-to-high power operations. This talk will focus on efforts to leverage the underlying time-dependent dynamics of plasma to investigate and influence thruster research and development. Prior years of study have developed techniques to uniquely represent the dynamics of such systems that have since been used to open a new way to test and operate plasma systems. Additional work has investigated the correlations between time-dependent measurements of these dynamics to develop digital twins, automate test processes with machine learning, inform design of experiments, and develop on-orbit system diagnostics. The talk will conclude with a look to the future as these tools are further applied both within the lab and potentially transitioned to on-orbit applications.
Speaker Bio
Dr. Christine Greve is a research engineer for the Air Force Research Laboratory at Edwards AFB. She received her Ph.D. in Aerospace Engineering from Texas A&M University under an NDSEG fellowship for her work in data-driven modeling of plasma-based systems. She now serves as the Electric Propulsion group lead with interests in high-power electric propulsion, machine learning, data-driven modeling, and novel plasma diagnostic techniques.
Beneath the Surface: Guidance, Navigation, and Control in the Ocean - From Theory to Lessons Learned
Abtract
Much of the underwater environment remains poorly mapped and observed, motivating the development of robotic systems that can safely and efficiently extend human reach beneath the surface. Yet the subsea domain imposes constraints beyond those typical in core robotics: no GPS, severely limited communication, challenging sensing, and strong environmental disturbances. This talk introduces the major classes of underwater robots and focuses on the guidance, navigation, and control pipeline that enables reliable operation in the ocean. I will review key background ideas, discuss underwater navigation in practice, and share take‑home lessons from practical experience spanning subsea operations, ROV prototyping, and work at the Monterey Bay Aquarium Research Institute (MBARI).
I will conclude by connecting these challenges to broader robotics themes, highlighting opportunities for collaboration between UCSC and MBARI.
Speaker Bio
Mauro Candeloro is a control engineer and postdoctoral researcher at the Monterey Bay Aquarium Research Institute (MBARI), where he develops planning and guidance methods for near-bottom seafloor mapping. He is a partner at Norwegian Subsea, where he previously served as Head of Products and Senior Technical Engineer, leading embedded firmware and sensor-fusion development for commercial motion reference units used in offshore navigation. He contributed to the development of the first consumer‑grade ROV prototype at BluEye Robotics. His experience spans research and industry field operations, including development of control systems for autonomous and remotely operated vehicles, as well as participation in offshore missions and Arctic expeditions. He earned his Ph.D. in Underwater Robotics Control from the Norwegian University of Science and Technology.
