INRC Forum Oct 4: Kapoho Point, Andrew Sornborger, Intel Q&A
Join us Tuesday, October 4, 2022 @ 9:00-10:00am PDT / 18:00-19:00 CET for an exciting INRC Forum!
Agenda
@Timothy Shea provides an update on Kapoho Point for the INRC and the Lava v0.5 release.
Research Talk by INRC member Andrew Sornborger of Los Alamos National Labs.
@Marcus Williams hosts first Intel Labs open-door community Q&A. Ask questions and get feedback from Lava developers.
Neural and Circuit Mechanisms for Neuromorphic Algorithms
Abstract: Over the past few years, we have been working on implementing a number of algorithms on neuromorphic chips. In order to do this, we have developed a range of techniques based on a simple portfolio of fundamental mechanisms. The workhorse mechanism is the synfire-gated synfire chain, which we use to control the flow of information on chip. When used in concert with other neural mechanisms and circuit structures, such as spike-timing-dependent plasticity, synaptic connectivity, and encoding schemes, we have been able to implement algorithms to copy synaptic weights from one circuit to another, to learn statistical processes, and, most recently, to construct a fully on-chip, spiking neuromorphic backpropagation algorithm. In this talk, I will discuss our neuromorphic programming framework and show, through examples, how it may be used to build algorithms of interest both for machine learning, as well as more standard algorithms that might be useful as modules in larger neural circuits.
Bio: Andrew Sornborger is a staff scientist at Los Alamos National Laboratory in the Information Sciences division. He worked in computational neuroscience before switching interests to neural and neuromorphic computation, which he has studied for a little over a decade. With Louis Tao (Peking University), he developed the concept of synfire-gated synfire chains (SGSCs) as a control framework for information processing and learning in neural systems. Based on this framework, he and collaborators have developed neural circuits for signal analysis, statistical learning, synaptic copy, and machine learning. He has also studied the theoretical underpinnings of SGSCs in terms of their bifurcation structure and robustness.
How to join:
INRC Members can find the meeting link on INRC Fall Forum Schedule. If you are interested to become a member, join the INRC.