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 neural networks (DNNs) have demonstrated their outperformance in varioussoftware systems, but also exhibit misbehavior and even result in irreversibledisasters. Therefore, it is crucial to identify the misbehavior of DNN-basedsoftware and improve DNNs' quality. Test input prioritization is one of themost appealing ways to guarantee DNNs' quality, which prioritizes test inputsso that more bug-revealing inputs can be identified earlier with limited timeand manual labeling efforts. However, the existing prioritization methods arestill limited from three aspects: certifiability, effectiveness, andgeneralizability. To overcome the challenges, we propose CertPri, a test inputprioritization technique designed based on a movement cost perspective of testinputs in DNNs' feature space. CertPri differs from previous works in three keyaspects: (1) certifiable: it provides a formal robustness guarantee for themovement cost; (2) effective: it leverages formally guaranteed movement coststo identify malicious bug-revealing inputs; and (3) generic: it can be appliedto various tasks, data, models, and scenarios. Extensive evaluations across 2tasks (i.e., classification and regression), 6 data forms, 4 model structures,and 2 scenarios (i.e., white-box and black-box) demonstrate CertPri's superiorperformance. For instance, it significantly improves 53.97% prioritizationeffectiveness on average compared with baselines. Its robustness andgeneralizability are 1.41~2.00 times and 1.33~3.39 times that of baselines onaverage, respectively.

CertPri: Certifiable Prioritization for Deep Neural Networks via Movement Cost in Feature Space