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

Unseen data conditions can inflict serious performance degradation on systemsrelying on supervised machine learning algorithms. Because data can often beunseen, and because traditional machine learning algorithms are trained in asupervised manner, unsupervised adaptation techniques must be used to adapt themodel to the unseen data conditions. However, unsupervised adaptation is oftenchallenging, as one must generate some hypothesis given a model and then usethat hypothesis to bootstrap the model to the unseen data conditions.Unfortunately, reliability of such hypotheses is often poor, given the mismatchbetween the training and testing datasets. In such cases, a model hypothesisconfidence measure enables performing data selection for the model adaptation.Underlying this approach is the fact that for unseen data conditions, datavariability is introduced to the model, which the model propagates to itsoutput decision, impacting decision reliability. In a fully connected network,this data variability is propagated as distortions from one layer to the next.This work aims to estimate the propagation of such distortion in the form ofnetwork activation entropy, which is measured over a short- time running windowon the activation from each neuron of a given hidden layer, and thesemeasurements are then used to compute summary entropy. This work demonstratesthat such an entropy measure can help to select data for unsupervised modeladaptation, resulting in performance gains in speech recognition tasks. Resultsfrom standard benchmark speech recognition tasks show that the proposedapproach can alleviate the performance degradation experienced under unseendata conditions by iteratively adapting the model to the unseen datas acousticcondition.

Leveraging Deep Neural Network Activation Entropy to cope with Unseen Data in Speech Recognition