Dresden 2020 – wissenschaftliches Programm

Die DPG-Frühjahrstagung in Dresden musste abgesagt werden! Lesen Sie mehr ...

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

SYNC: Symposium Advanced neuromorphic computing hardware: Towards efficient machine learning

SYNC 1: Advanced neuromorphic computing hardware: Towards efficient machine learning

SYNC 1.2: Hauptvortrag

Montag, 16. März 2020, 10:00–10:30, HSZ 01

Metal-oxide resistance switching memory devices as artificial synapses for brain-inspired computing — •Sabina Spiga — CNR-IMM, via C. Olivetti 2, 20864 Agrate Brianza (MB), Italy

Memristive devices have been receiving an increasing interest for a wide range of applications, such as storage class memory, non-volatile logic switch, in-memory computing and neuromorphic computing. In particular, in bio-inspired systems, memristive devices can act as dispersed memory elements mimicking synapses in nervous systems, or as stochastic and non-linear elements in neuronal units. Among the proposed technologies, oxide-based resistance switching memory devices (RRAM) are based on redox reactions and electrochemical phenomena in oxides and are very promising because of low power consumption, fast switching times, scalability down to nm scale and CMOS compatibility. In our work, we focus on the switching dynamics of HfO2-based RRAMs and on their implementation as electronic synapses in spiking neural networks (SNN). The conductance of the RRAM devices can be tuned in an analog fashion over several states by using proper programming algorithms and by engineering the material stack. The device dynamics is used to emulate the biological potentiation (conductance increase) and depression (conductance decrease) processes, over several cycles. Moreover, the conductance value update can be achieved by a spike timing and rate dependent plasticity mechanism, which is demonstrated at hardware level, and that is exploited as learning rule in a SNN.