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

Given a small training data set and a learning algorithm, how much more datais necessary to reach a target validation or test performance? This question isof critical importance in applications such as autonomous driving or medicalimaging where collecting data is expensive and time-consuming. Overestimatingor underestimating data requirements incurs substantial costs that could beavoided with an adequate budget. Prior work on neural scaling laws suggest thatthe power-law function can fit the validation performance curve and extrapolateit to larger data set sizes. We find that this does not immediately translateto the more difficult downstream task of estimating the required data set sizeto meet a target performance. In this work, we consider a broad class ofcomputer vision tasks and systematically investigate a family of functions thatgeneralize the power-law function to allow for better estimation of datarequirements. Finally, we show that incorporating a tuned correction factor andcollecting over multiple rounds significantly improves the performance of thedata estimators. Using our guidelines, practitioners can accurately estimatedata requirements of machine learning systems to gain savings in bothdevelopment time and data acquisition costs.

How Much More Data Do I Need? Estimating Requirements for Downstream Tasks