Dresden 2017 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

DS: Fachverband Dünne Schichten

DS 40: Focussed Session: Memristive Devices for Neuronal Systems II

DS 40.1: Topical Talk

Thursday, March 23, 2017, 15:00–15:30, CHE 89

Brain-inspired neurocomputing with memristive synapses — •Daniele Ielmini — Dipartimento di Elettronica, Informazione e Bioingegneria, Politecnico di Milano, Italy

As Moore's law of CMOS downscaling slows down, research on alternative computing devices and architectures receives an increasing interest. Among these, memcomputing and neurocomputing combine a relatively simple concept (e.g., analog/digital circuits, CMOS compatible materials, room operation operation) with attractive functionalities, such as reconfigurability, ultra-low power operation, and brain-like computing capabilities including learning and inference. Merging memcomputing and neurocomputing may result in novel conceptual schemes with high energy efficiency, high speed, excellent scalability, and brain-like computing capabilities.

This talk will show recent advances on neuro-computing using memristive synapses made of resistive switching memory (RRAM) devices. Hybrid CMOS/memristive synapses capable of spike-timing dependent plasticity (STDP) and spike-rate dependent plasticity (SRDP) will be experimentally demonstrated. Pattern learning will be demonstrated by a feedforward neural network with up to 4x4 input images, 2 post-synaptic firing neurons, handling both static and dynamic moving images. The prospects of such new achievements for brain-inspired computing will be discussed.