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

Autoencoder-based learning has emerged as a staple for discipliningrepresentations in unsupervised and semi-supervised settings. This paperanalyzes a framework for improving generalization in a purely supervisedsetting, where the target space is high-dimensional. We motivate and formalizethe general framework of target-embedding autoencoders (TEA) for supervisedprediction, learning intermediate latent representations jointly optimized tobe both predictable from features as well as predictive of targets---encodingthe prior that variations in targets are driven by a compact set of underlyingfactors. As our theoretical contribution, we provide a guarantee ofgeneralization for linear TEAs by demonstrating uniform stability, interpretingthe benefit of the auxiliary reconstruction task as a form of regularization.As our empirical contribution, we extend validation of this approach beyondexisting static classification applications to multivariate sequenceforecasting, verifying their advantage on both linear and nonlinear recurrentarchitectures---thereby underscoring the further generality of this frameworkbeyond feedforward instantiations.

Target-Embedding Autoencoders for Supervised Representation Learning