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

Existing deep learning approaches for learning visual features tend tooverlearn and extract more information than what is required for the task athand. From a privacy preservation perspective, the input visual information isnot protected from the model; enabling the model to become more intelligentthan it is trained to be. Current approaches for suppressing additional tasklearning assume the presence of ground truth labels for the tasks to besuppressed during training time. In this research, we propose a three-foldnovel contribution: (i) a model-agnostic solution for reducing modeloverlearning by suppressing all the unknown tasks, (ii) a novel metric tomeasure the trust score of a trained deep learning model, and (iii) a simulatedbenchmark dataset, PreserveTask, having five different fundamental imageclassification tasks to study the generalization nature of models. In the firstset of experiments, we learn disentangled representations and suppressoverlearning of five popular deep learning models: VGG16, VGG19, Inception-v1,MobileNet, and DenseNet on PreserverTask dataset. Additionally, we show resultsof our framework on color-MNIST dataset and practical applications of faceattribute preservation in Diversity in Faces (DiF) and IMDB-Wiki dataset.

Reducing Overlearning through Disentangled Representations by Suppressing Unknown Tasks