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

Real-life applications, heavily relying on machine learning, such as dialogsystems, demand out-of-domain detection methods. Intent classification modelsshould be equipped with a mechanism to distinguish seen intents from unseenones so that the dialog agent is capable of rejecting the latter and avoidingundesired behavior. However, despite increasing attention paid to the task, thebest practices for out-of-domain intent detection have not yet been fullyestablished.  This paper conducts a thorough comparison of out-of-domain intent detectionmethods. We prioritize the methods, not requiring access to out-of-domain dataduring training, gathering of which is extremely time- and labor-consuming dueto lexical and stylistic variation of user utterances. We evaluate multiplecontextual encoders and methods, proven to be efficient, on three standarddatasets for intent classification, expanded with out-of-domain utterances. Ourmain findings show that fine-tuning Transformer-based encoders on in-domaindata leads to superior results. Mahalanobis distance, together with utterancerepresentations, derived from Transformer-based encoders, outperforms othermethods by a wide margin and establishes new state-of-the-art results for alldatasets.  The broader analysis shows that the reason for success lies in the fact thatthe fine-tuned Transformer is capable of constructing homogeneousrepresentations of in-domain utterances, revealing geometrical disparity to outof domain utterances. In turn, the Mahalanobis distance captures this disparityeasily.  The code is available in our GitHub repo:https://github.com/huawei-noah/noah-research/tree/master/Maha_OOD .

Revisiting Mahalanobis Distance for Transformer-Based Out-of-Domain Detection