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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Deep Learning models possess two key traits that, in combination, make theiruse in the real world a risky prospect. One, they do not typically generalizewell outside of the distribution for which they were trained, and two, theytend to exhibit confident behavior regardless of whether or not they areproducing meaningful outputs. While Deep Learning possesses immense power tosolve realistic, high-dimensional problems, these traits in concert make itdifficult to have confidence in their real-world applications. To overcome thisdifficulty, the task of Out-Of-Distribution (OOD) Detection has been defined,to determine when a model has received an input from outside of thedistribution for which it is trained to operate.  This paper introduces and examines a novel methodology, DOODLER, for OODDetection, which directly leverages the traits which result in its necessity.By training a Variational Auto-Encoder (VAE) on the same data as another DeepLearning model, the VAE learns to accurately reconstruct In-Distribution (ID)inputs, but not to reconstruct OOD inputs, meaning that its failure state canbe used to perform OOD Detection. Unlike other work in the area, DOODLERrequires only very weak assumptions about the existence of an OOD dataset,allowing for more realistic application. DOODLER also enables pixel-wisesegmentations of input images by OOD likelihood, and experimental results showthat it matches or outperforms methodologies that operate under the sameconstraints.

DOODLER: Determining Out-Of-Distribution Likelihood from Encoder Reconstructions