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/

Zendy Plus

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

Zendy Tools

Zendy Open

To adopt convolutional neural networks (CNN) for a range ofresource-constrained targets, it is necessary to compress the CNN models byperforming quantization, whereby precision representation is converted to alower bit representation. To overcome problems such as sensitivity of thetraining dataset, high computational requirements, and large time consumption,post-training quantization methods that do not require retraining have beenproposed. In addition, to compensate for the accuracy drop without retraining,previous studies on post-training quantization have proposed severalcomplementary methods: calibration, schemes, clipping, granularity, andmixed-precision. To generate a quantized model with minimal error, it isnecessary to study all possible combinations of the methods because each ofthem is complementary and the CNN models have different characteristics.However, an exhaustive or a heuristic search is either too time-consuming orsuboptimal. To overcome this challenge, we propose an auto-tuner known asQuantune, which builds a gradient tree boosting model to accelerate the searchfor the configurations of quantization and reduce the quantization error. Weevaluate and compare Quantune with the random, grid, and genetic algorithms.The experimental results show that Quantune reduces the search time forquantization by approximately 36.5x with an accuracy loss of 0.07 ~ 0.65%across six CNN models, including the fragile ones (MobileNet, SqueezeNet, andShuffleNet). To support multiple targets and adopt continuously evolvingquantization works, Quantune is implemented on a full-fledged compiler for deeplearning as an open-sourced project.

Quantune: Post-training Quantization of Convolutional Neural Networks using Extreme Gradient Boosting for Fast Deployment