Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
SYCC: Symposium Identifying Optimal Physical Implementations for beyond von Neumann Computing Concepts
SYCC 1: Identifying optimal physical implementations of non-conventional computing
SYCC 1.3: Hauptvortrag
Freitag, 5. April 2019, 10:30–11:00, H1
Neuromorphic computing with spintronic nano-oscillators — •Philippe Talatchian1, Miguel Romera1, Sumito Tsunegi2, Flavio Abreu Araujo1, Vincent Cros1, Paolo Bortolotti1, Juan Trastoy1, Kay Yakushiji2, Akio Fukushima2, Hitoshi Kubota2, Shinji Yuasa2, Maxence Ernoult1,3, Damir Vodenicarevic3, Tifenn Hirtzlin3, Nicolas Locatelli3, Damien Querlioz3, and Julie Grollier1 — 1Unité Mixte de Physique, CNRS, Thales, Université Paris-Sud, Palaiseau, France — 2Spintronics Research Center (AIST), Tsukuba, Ibaraki, Japan — 3C2N, CNRS, Université Paris-Sud, Orsay, France
Biological neurons emit periodic electrical spikes and can synchronize their rhythmic activity. Inspired from these features, it would be attractive to implement oscillatory neural networks for computing in hardware, with nanoscale oscillators capable of synchronization. However, despite numerous proposals, there is today no demonstration of brain-inspired computing with coupled nano-oscillators. One difficulty is that training these networks requires tuning the coupling between oscillators. Here we show experimentally that through their high frequency tunability, spintronic nano-oscillators can learn to perform pattern recognition. We train a network of four coupled spin-torque nano-oscillators to recognize spoken vowels with an experimental recognition rate of 88 percents [1]. These results open new paths towards highly energy efficient bio-inspired computing on-chip based on non-linear nano-devices.
[1] M. Romera, P. Talatchian et al, Nature. Vol 563, p.230-234, 2018.