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

Deep Learning (DL) has recently achieved tremendous success. A variety of DLframeworks and platforms play a key role to catalyze such progress. However,the differences in architecture designs and implementations of existingframeworks and platforms bring new challenges for DL software development anddeployment. Till now, there is no study on how various mainstream frameworksand platforms influence both DL software development and deployment inpractice. To fill this gap, we take the first step towards understanding howthe most widely-used DL frameworks and platforms support the DL softwaredevelopment and deployment. We conduct a systematic study on these frameworksand platforms by using two types of DNN architectures and three populardatasets. (1) For development process, we investigate the prediction accuracyunder the same runtime training configuration or same model weights/biases. Wealso study the adversarial robustness of trained models by leveraging theexisting adversarial attack techniques. The experimental results show that thecomputing differences across frameworks could result in an obvious predictionaccuracy decline, which should draw the attention of DL developers. (2) Fordeployment process, we investigate the prediction accuracy and performance(refers to time cost and memory consumption) when the trained models aremigrated/quantized from PC to real mobile devices and web browsers. The DLplatform study unveils that the migration and quantization still suffer fromcompatibility and reliability issues. Meanwhile, we find several DL softwarebugs by using the results as a benchmark. We further validate the resultsthrough bug confirmation from stakeholders and industrial positive feedback tohighlight the implications of our study. Through our study, we summarizepractical guidelines, identify challenges and pinpoint new research directions.

An Empirical Study towards Characterizing Deep Learning Development and Deployment across Different Frameworks and Platforms