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

Classification neural networks fail to detect inputs that do not fall insidethe classes they have been trained for. Runtime monitoring techniques on theneuron activation pattern can be used to detect such inputs. We present anapproach for monitoring classification systems via data abstraction. Dataabstraction relies on the notion of box with a resolution. Box-basedabstraction consists in representing a set of values by its minimal and maximalvalues in each dimension. We augment boxes with a notion of resolution anddefine their clustering coverage, which is intuitively a quantitative metricthat indicates the abstraction quality. This allows studying the effect ofdifferent clustering parameters on the constructed boxes and estimating aninterval of sub-optimal parameters. Moreover, we automatically constructmonitors that leverage both the correct and incorrect behaviors of a system.This allows checking the size of the monitor abstractions and analyzing theseparability of the network. Monitors are obtained by combining thesub-monitors of each class of the system placed at some selected layers. Ourexperiments demonstrate the effectiveness of our clustering coverage estimationand show how to assess the effectiveness and precision of monitors according tothe selected clustering parameter and monitored layers.

Customizable Reference Runtime Monitoring of Neural Networks using Resolution Boxes