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

Autoencoders are unsupervised deep learning models used for learningrepresentations. In literature, autoencoders have shown to perform well on avariety of tasks spread across multiple domains, thereby establishingwidespread applicability. Typically, an autoencoder is trained to generate amodel that minimizes the reconstruction error between the input and thereconstructed output, computed in terms of the Euclidean distance. While thiscan be useful for applications related to unsupervised reconstruction, it maynot be optimal for classification. In this paper, we propose a novel SupervisedCOSMOS Autoencoder which utilizes a multi-objective loss function to learnrepresentations that simultaneously encode the (i) "similarity" between theinput and reconstructed vectors in terms of their direction, (ii)"distribution" of pixel values of the reconstruction with respect to the inputsample, while also incorporating (iii) "discriminability" in the featurelearning pipeline. The proposed autoencoder model incorporates a Cosinesimilarity and Mahalanobis distance based loss function, along with supervisionvia Mutual Information based loss. Detailed analysis of each component of theproposed model motivates its applicability for feature learning in differentclassification tasks. The efficacy of Supervised COSMOS autoencoder isdemonstrated via extensive experimental evaluations on different imagedatasets. The proposed model outperforms existing algorithms on MNIST,CIFAR-10, and SVHN databases. It also yields state-of-the-art results onCelebA, LFWA, Adience, and IJB-A databases for attribute prediction and facerecognition, respectively.

Supervised COSMOS Autoencoder: Learning Beyond the Euclidean Loss!