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

Recent trends show recognition accuracy increasing even more profoundly.Inference process of Deep Convolutional Neural Networks (DCNN) has a largenumber of parameters, requires a large amount of computation, and can be veryslow. The large number of parameters also require large amounts of memory. Thisis resulting in increasingly long computation times and large model sizes. Toimplement mobile and other low performance devices incorporating DCNN, modelsizes must be compressed and computation must be accelerated. To that end, thispaper proposes Binary-decomposed DCNN, which resolves these issues without theneed for retraining. Our method replaces real-valued inner-product computationswith binary inner-product computations in existing network models to acceleratecomputation of inference and decrease model size without the need forretraining. Binary computations can be done at high speed using logicaloperators such as XOR and AND, together with bit counting. In tests usingAlexNet with the ImageNet classification task, speed increased by a factor of1.79, models were compressed by approximately 80%, and increase in error ratewas limited to 1.20%. With VGG-16, speed increased by a factor of 2.07, modelsizes decreased by 81%, and error increased by only 2.16%.

Binary-decomposed DCNN for accelerating computation and compressing model without retraining