Robert Ivans, Idaho National Laboratory
“Much ado about nothing: hardware that learns without software"
Thursday, November 19 – 4:00 to 5:00 p.m.
Meeting URL:
https://bluejeans.com/159740539Meeting ID: 159 740 539
Abstract: The human brain is composed of billions of neurons connected through trillions of synapses to form a massively parallel, fault tolerant, and energy efficient pattern matching machine. These features are the envy of all computer engineers. The goal of neuromorphic design is to create circuitry with these desirable characteristics by mimicking biology and using physics-based computation, and biologically inspired learning rules such as spike-timing-dependent plasticity, to perform tasks. One such task is spatiotemporal pattern recognition. This presentation will be a high-level introduction to spiking neuromorphic spatiotemporal pattern recognition from the ground up. Topics will include biology, history, and current research.
Bio: Robert Ivans is a PhD candidate out of Boise State University, under the guidance of Kurtis Cantley, where his research uses spiking neuromorphic designs to perform tasks such as unsupervised spatiotemporal pattern detection. In his most recent paper, published in the IEEE Transactions on Neural Networks and Learning Systems (impact factor=8.79), he created a model for R(t) elements and showed how they can be used to facilitate STDP in neuromorphic designs with memristive synapses. Robert was one of twelve students selected to be in the third cohort for the prestigious INL Graduate Fellowship in 2019, and has recently moved to Ammon, Idaho where he resides happily with his wife, Kristen.