Learning and computation in disordered networks of memristors: theory and experiments
Lead Research Organisation:
University of the West of England
Department Name: Faculty of Environment and Technology
Abstract
Memristor (memory resistor) is a device whose resistance changes depending on the polarity and magnitude of a voltage applied to the device's terminals and the duration of this voltage's application. The memristor is a non-volatile memory because the specific resistance is retained until the application of another voltage. A memristor implements a material version of Boolean logic and thus any logical circuit can be constructed from memristors. We propose to fabricate in laboratory experiments an adaptive, self-organized disordered network of memristors. This practical fabrication will be backed up by rigorous computer simulation experiments. The memristor network is comprised of a conglomerate of conductive polymer fibres interspersed with particles of solid electrolyte. The conglomerate is placed on a matrix of micro-electrodes capable of recording voltage and generating current sources and sinks. Machine learning techniques will be applied in order to design logical schemes and basic arithmetical circuits.
Publications
EROKHIN V
(2012)
ORGANIC MEMRISTOR DEVICES FOR LOGIC ELEMENTS WITH MEMORY
in International Journal of Bifurcation and Chaos
Gale E
(2012)
Memristor-based information gathering approaches, both ant-inspired and hypothetical
in Nano Communication Networks
Gale E
(2013)
Observation, Characterization and Modeling of Memristor Current Spikes
in Applied Mathematics & Information Sciences
Gale E
(2014)
Drop-coated titanium dioxide memristors
in Materials Chemistry and Physics
Gale Ella
(2011)
The Memory-Conservation Theory of Memristance
in arXiv e-prints
Howard D
(2015)
Evolving unipolar memristor spiking neural networks
in Connection Science
HOWARD G
(2013)
A SPICE MODEL OF THE PEO-PANI MEMRISTOR
in International Journal of Bifurcation and Chaos
Howard G
(2012)
Genetic Programming
| Description | We developed a spiking neuro evolutionary system which implements memristors as plastic connections, i.e., where weights can vary during a trail. The evolutionary design process exploits parameter self-adaption and variable topologies, allowing the number of neurons, connection weights, and inter-neural connectivity pattern to emerge. By comparing two phenomenological real-world memristor implementations with networks comprised of: 1) linear resistors, and 2) constant-valued connections, we demonstrate that this approach allows the evolution of networks of appropriate complexity to emerge whilst exploiting the memristive properties of the connections to reduce learning time. We extend this approach to allow for heterogeneous mixtures of memristors within the networks; our approach provides an in-depth analysis of network structure. Or networks are evaluated on simulated robotic navigation tasks; results demonstrate that memristive plasticity enables higher performance than constant-weighted connections in both static and dynamic reward scenarios, and that mixtures of memristive elements provide performance advantages when compared to homogenous memristive networks. Research publications can be found on arxiv.org and scholar.google.co.uk |
| Exploitation Route | In the design of novel learning computer hardware |
| Sectors | Aerospace, Defence and Marine,Digital/Communication/Information Technologies (including Software),Electronics |
| Description | The findings were used to establish a new paradigm of computing using thin film and polymer based resistors with memory. Discoveries of spiking behaviour in the memristors led to novel types of sequential logics. |
| First Year Of Impact | 2013 |
| Sector | Digital/Communication/Information Technologies (including Software),Electronics |
| Impact Types | Economic |