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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

AutoAugment has sparked an interest in automated augmentation methods fordeep learning models. These methods estimate image transformation policies fortrain data that improve generalization to test data. While recent papersevolved in the direction of decreasing policy search complexity, we show thatthose methods are not robust when applied to biased and noisy data. To overcomethese limitations, we reformulate AutoAugment as a generalized automateddataset optimization (AutoDO) task that minimizes the distribution shiftbetween test data and distorted train dataset. In our AutoDO model, weexplicitly estimate a set of per-point hyperparameters to flexibly changedistribution of train data. In particular, we include hyperparameters foraugmentation, loss weights, and soft-labels that are jointly estimated usingimplicit differentiation. We develop a theoretical probabilistic interpretationof this framework using Fisher information and show that its complexity scaleslinearly with the dataset size. Our experiments on SVHN, CIFAR-10/100, andImageNet classification show up to 9.3% improvement for biased datasets withlabel noise compared to prior methods and, importantly, up to 36.6% gain forunderrepresented SVHN classes.

AutoDO: Robust AutoAugment for Biased Data with Label Noise via Scalable Probabilistic Implicit Differentiation