English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Memristive nanodevices offer new frontiers for computing systems that unitearithmetic and memory operations on-chip. Here, we explore the integration ofelectrochemical metallization cell (ECM) nanodevices with tunable filamentaryswitching in nanoscale learning systems. Such devices offer a naturaltransition between short-term plasticity (STP) and long-term plasticity (LTP).In this work, we show that this property can be exploited to efficiently solvenoisy classification tasks. A single crossbar learning scheme is firstintroduced and evaluated. Perfect classification is possible only for simpleinput patterns, within critical timing parameters, and when device variabilityis weak. To overcome these limitations, a dual-crossbar learning system partlyinspired by the extreme learning machine (ELM) approach is then introduced.This approach outperforms a conventional ELM-inspired system when the firstlayer is imprinted before training and testing, and especially so whenvariability in device timing evolution is considered: variability is thereforetransformed from an issue to a feature. In attempting to classify the MNISTdatabase under the same conditions, conventional ELM obtains 84%classification, the imprinted, uniform device system obtains 88%classification, and the imprinted, variable device system reaches 92%classification. We discuss benefits and drawbacks of both systems in terms ofenergy, complexity, area imprint, and speed. All these results highlight thattuning and exploiting intrinsic device timing parameters may be of centralinterest to future bio-inspired approximate computing systems.

Exploiting the Short-term to Long-term Plasticity Transition in Memristive Nanodevice Learning Architectures