CPOS Seminar: "Building Energy-Efficient AI: Neuromorphic Hardware Challenges and Materials Opportunities"

Speaker: Sai Sukruth Bezugam, PhD Student. Strukovs Research Group, Department of ECE, UC Santa Barbara

How can we design next-generation AI systems that meet stringent energy requirements while maintaining high performance? Neuromorphic computing provides a potential answer by emulating key principles of biological neural networks, such as massive parallelism and in-memory processing, leading to significant reductions in power consumption compared to conventional architectures. In this seminar, I will discuss the core concepts underlying neuromorphic hardware, focusing on how artificial synapses and neurons should ideally function to enable energy-efficient operation. I will also highlight the principal challenges—such as linearity, variability, and endurance—that must be addressed to translate these devices into practical systems. Additionally, I will consider how polymeric and organic materials, with their unique and tunable properties, may be investigated for their potential to address these challenges. By examining neuromorphic hardware from both a design and materials perspective, this talk aims to stimulate interdisciplinary discussion on realizing low-power, scalable AI technologies.