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

We present a method to compute the derivative of a learning task with respectto a dataset. A learning task is a function from a training set to thevalidation error, which can be represented by a trained deep neural network(DNN). The "dataset derivative" is a linear operator, computed around thetrained model, that informs how perturbations of the weight of each trainingsample affect the validation error, usually computed on a separate validationdataset. Our method, DIVA (Differentiable Validation) hinges on a closed-formdifferentiable expression of the leave-one-out cross-validation error around apre-trained DNN. Such expression constitutes the dataset derivative. DIVA couldbe used for dataset auto-curation, for example removing samples with faultyannotations, augmenting a dataset with additional relevant samples, orrebalancing. More generally, DIVA can be used to optimize the dataset, alongwith the parameters of the model, as part of the training process without theneed for a separate validation dataset, unlike bi-level optimization methodscustomary in AutoML. To illustrate the flexibility of DIVA, we reportexperiments on sample auto-curation tasks such as outlier rejection, datasetextension, and automatic aggregation of multi-modal data.

DIVA: Dataset Derivative of a Learning Task