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

Transfer learning has been recently popularized as a data-efficientalternative to training models from scratch, in particular for computer visiontasks where it provides a remarkably solid baseline. The emergence of richmodel repositories, such as TensorFlow Hub, enables the practitioners andresearchers to unleash the potential of these models across a wide range ofdownstream tasks. As these repositories keep growing exponentially, efficientlyselecting a good model for the task at hand becomes paramount. We provide aformalization of this problem through a familiar notion of regret and introducethe predominant strategies, namely task-agnostic (e.g. ranking models by theirImageNet performance) and task-aware search strategies (such as linear or kNNevaluation). We conduct a large-scale empirical study and show that bothtask-agnostic and task-aware methods can yield high regret. We then propose asimple and computationally efficient hybrid search strategy which outperformsthe existing approaches. We highlight the practical benefits of the proposedsolution on a set of 19 diverse vision tasks.

Which Model to Transfer? Finding the Needle in the Growing Haystack