Paxon Frady (Intel Labs) - Breaking the Surface: progress and challenges with Intel's Loihi 2
Presenting...
Recording: https://youtu.be/qaxWAmamxd0
Abstract
Neuromorphic computing takes inspiration from the brain to rethink the fundamentals of computer architecture. Intel’s most recent neuromorphic chip, Loihi 2, is an asynchronous spiking hardware with on-chip learning. The promise of the architecture is low-latency parallel computing at very low energy. While Loihi 2 still uses conventional transistors and fabrication, its architecture requires a new paradigm for programming and algorithm design. I will discuss the basics of Loihi 2 and progress in applications, such as deep-learning and QUBO solvers, as well as our efforts to develop a new programming framework for distributed parallel hardware. While much of our focus internally has been to develop AI models, there is a lot of promise for Loihi 2 applications in other domains, such as signal processing, embodied cognition, and physics/fluid dynamics simulations. Through our academic community, the INRC, we provide remote access to the hardware, and we have an open-source software library, lava.
Speaker Bio
E. Paxon Frady received his B.S. at the California Institute of Technology with a double major in Computation and Neural Systems and Business, Economics and Management. He received a Ph.D. in Neuroscience at the University of California San Diego, with specialization in Computational Neuroscience. He was a Postdoctoral Researcher at the Redwood Center for Theoretical Neuroscience at UC Berkeley. Currently, he is a research scientist in the Neuromorphic Computing Lab at Intel.