Thierry Tambe, Assistant Professor, Department of Electrical Engineering

Abstract: The unabated pursuit for omniscient and omnipotent AI is levying hefty latency, memory, and energy taxes at all computing scales. Given the limits of CMOS scaling, my research is building a heterogeneity of solutions cooptimized across the application’s algorithm, memory subsystem, hardware architecture, and silicon stack to generate breakthrough advances in arithmetic performance, compute flexibility, and energy efficiency for on-chip deep learning. We have developed several edge AI chips to validate these solutions. In this talk, I will be highlighting key learnings gleaned from my experience working on these domain-specific chip projects, loosely distilled into five categories: 1) number systems, 2) sparsity, 3) memory, 4) power management, and 5) chip design productivity. I will point out some notable architectural trends and offer perspectives on next-generation VLSI systems for generative AI workloads.

Zerina Kapetanovic, Assistant Professor, Department of Electrical Engineering

Abstract: The Internet of Things (IoT) plays a critical role in connecting society to the digital world. Billions of IoT devices are used today to enable applications from smart homes and cities to digital healthcare and smart agriculture. While we have seen significant benefits, to fully realize the potential of these systems we need to address a key challenge: enabling power and connectivity for resource-constrained environments. How can we collect data in remote forests, farms, or even oceans where there is no Internet connectivity or access to power?

In this talk, I will present two examples of my research that address this challenge. First, I will present FarmBeats, a system that enables seamless data collection by using new technologies such as TV White Spaces for communication and low-power sensors for data collection. FarmBeats is designed to enable data-driven agriculture techniques for farms that are located in areas that lack power and connectivity infrastructure. I have deployed the FarmBeats system across the United States and Europe, and we have demonstrated that the system can provide significant benefits for farms in the form of cost savings and productivity. Finally, I will present the design and implementation of a low-power passive wireless communication system that does not rely on ambient or generated RF signals and instead modulates Johnson (thermal) noise.