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

It remains a challenge to run Deep Learning in devices with stringent powerbudget in the Internet-of-Things. This paper presents a low-power acceleratorfor processing Deep Neural Networks in the embedded devices. The powerreduction is realized by avoiding multiplications of near-zero valued data. Thenear-zero approximation and a dedicated Near-Zero Approximation Unit (NZAU) areproposed to predict and skip the near-zero multiplications under certainthresholds. Compared with skipping zero-valued computations, our designachieves 1.92X and 1.51X further reduction of the total multiplications inLeNet-5 and Alexnet respectively, with negligible lose of accuracy. In theproposed accelerator, 256 multipliers are grouped into 16 independentProcessing Lanes (PL) to support up to 16 neuron activations simultaneously.With the help of data pre-processing and buffering in each PL, multipliers canbe clock-gated in most of the time even the data is excessively streaming in.Designed and simulated in UMC 65 nm process, the accelerator operating at 500MHz is $>$ 4X faster than the mobile GPU Tegra K1 in processing thefully-connected layer FC8 of Alexnet, while consuming 717X less energy.

A Low-Power Accelerator for Deep Neural Networks with Enlarged Near-Zero Sparsity