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

We present a novel technique, called Term Revealing (TR), for furtheringquantization at run time for improved performance of Deep Neural Networks(DNNs) already quantized with conventional quantization methods. TR operates onpower-of-two terms in binary expressions of values. In computing a dot-productcomputation, TR dynamically selects a fixed number of largest terms to use fromthe values of the two vectors in the dot product. By exploiting normal-likeweight and data distributions typically present in DNNs, TR has a minimalimpact on DNN model performance (i.e., accuracy or perplexity). We use TR tofacilitate tightly synchronized processor arrays, such as systolic arrays, forefficient parallel processing. We show an FPGA implementation that can use asmall number of control bits to switch between conventional quantization andTR-enabled quantization with a negligible delay. To enhance TR efficiencyfurther, we use a signed digit representation (SDR), as opposed to classicbinary encoding with only nonnegative power-of-two terms. To perform conversionfrom binary to SDR, we develop an efficient encoding method called HESE (HybridEncoding for Signed Expressions) that can be performed in one pass looking atonly two bits at a time. We evaluate TR with HESE encoded values on an MLP forMNIST, multiple CNNs for ImageNet, and an LSTM for Wikitext-2, and showsignificant reductions in inference computations (between 3-10x) compared toconventional quantization for the same level of model performance.

Term Revealing: Furthering Quantization at Run Time on Quantized DNNs