BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:An organic artificial synapse for low-energy neuromorphic computin
 g - Yoeri van de Burgt (Eindhoven University of Technology)
DTSTART:20171107T130000Z
DTEND:20171107T140000Z
UID:TALK95254@talks.cam.ac.uk
CONTACT:Yoeri van de Burgt
DESCRIPTION:The widely anticipated end to Moore’s law and the growing de
 mand for low power computing systems capable of learning\, image recogniti
 on and real-time analysis of large streams of unstructured data has spurre
 d intense interest in neural algorithms for brain-inspired computing. Toda
 y these neural networks can not only translate languages and classify imag
 es but are also successfully implemented in recognizing diseases\, identif
 y jaywalkers\, predict optimised routes and unlock your phone. Still\, the
  volatility\, high supply voltages\, and number of transistors required pe
 r synapse significantly complicate the path for CMOS-based architectures t
 o achieve the extreme interconnectivity\, information density\, and energy
  efficiency of the brain.\n\nAlternatively\, two-terminal tunable resistan
 ce elements (memristors) based on filament forming metal oxides (FFMOs) or
  phase change memory (PCM) materials have been demonstrated to function as
  non-volatile memory that can emulate synaptic functions.  However\, memri
 stors demonstrated to date suffer from excessive write noise\, write nonli
 nearity\, and high write voltages. Reducing the noise and lowering the swi
 tching-voltage without compromising long-term data retention has proven di
 fficult.\n\nHere we present an organic neuromorphic device operating with 
 a fundamentally different mechanism from existing memristors\, based on th
 e electrochemical doping of a conjugated polymer . Controlled ion injectio
 n into the bulk of the material facilitates modification of the conductanc
 e or synaptic weight in a near-analogue fashion over a wide range. The dev
 ice switches at low energy and voltage\, displays >500 distinct\, non-vola
 tile conductance states and achieves high classification accuracy when imp
 lemented in neural network simulations. We also demonstrate that plastic E
 NODEs can be entirely fabricated on flexible substrates\, introducing neur
 omorphic computing to large-area flexible electronics and opening up possi
 bilities in brain-machine interfacing\, adaptive learning of artificial or
 gans\, such as “smart skins”\, and laying the foundation for 3D manufa
 cturing of highly interconnected device networks.
LOCATION:Electrical Engineering Division Seminar Room\, 9 J J Thomson Aven
 ue
END:VEVENT
END:VCALENDAR