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

The learning of hierarchical representations for image classification hasexperienced an impressive series of successes due in part to the availabilityof large-scale labeled data for training. On the other hand, the trainedclassifiers have traditionally been evaluated on small and fixed sets of testimages, which are deemed to be extremely sparsely distributed in the space ofall natural images. It is thus questionable whether recent performanceimprovements on the excessively re-used test sets generalize to real-worldnatural images with much richer content variations. Inspired by efficientstimulus selection for testing perceptual models in psychophysical andphysiological studies, we present an alternative framework for comparing imageclassifiers, which we name the MAximum Discrepancy (MAD) competition. Ratherthan comparing image classifiers using fixed test images, we adaptively samplea small test set from an arbitrarily large corpus of unlabeled images so as tomaximize the discrepancies between the classifiers, measured by the distanceover WordNet hierarchy. Human labeling on the resulting model-dependent imagesets reveals the relative performance of the competing classifiers, andprovides useful insights on potential ways to improve them. We report the MADcompetition results of eleven ImageNet classifiers while noting that theframework is readily extensible and cost-effective to add future classifiersinto the competition. Codes can be found at https://github.com/TAMU-VITA/MAD.

I Am Going MAD: Maximum Discrepancy Competition for Comparing Classifiers Adaptively