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

There is widespread interest in emerging technologies, especially resistivecrossbars for accelerating Deep Neural Networks (DNNs). Resistive crossbarsoffer a highly-parallel and efficient matrix-vector-multiplication (MVM)operation. MVM being the most dominant operation in DNNs makes crossbarsideally suited. However, various sources of device and circuit non-idealitieslead to errors in the MVM output, thereby reducing DNN accuracy. Towards thatend, we propose crossbar re-mapping strategies to mitigate line-resistanceinduced accuracy degradation in DNNs, without having to re-train the learnedweights, unlike most prior works. Line-resistances degrade the voltage levelsalong the crossbar columns, thereby inducing more errors at the columns awayfrom the drivers. We rank the DNN weights and kernels based on a sensitivityanalysis, and re-arrange the columns such that the most sensitive kernels aremapped closer to the drivers, thereby minimizing the impact of errors on theoverall accuracy. We propose two algorithms $-$ static remapping strategy (SRS)and dynamic remapping strategy (DRS), to optimize the crossbar re-arrangementof a pre-trained DNN. We demonstrate the benefits of our approach on a standardVGG16 network trained using CIFAR10 dataset. Our results show that SRS and DRSlimit the accuracy degradation to 2.9\% and 2.1\%, respectively, compared to a5.6\% drop from an as it is mapping of weights and kernels to crossbars. Webelieve this work brings an additional aspect for optimization, which can beused in tandem with existing mitigation techniques, such as in-situcompensation, technology aware training and re-training approaches, to enhancesystem performance.

X-CHANGR: Changing Memristive Crossbar Mapping for Mitigating Line-Resistance Induced Accuracy Degradation in Deep Neural Networks