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

There exists a distribution discrepancy between training and testing, in theway images are fed to modern CNNs. Recent work tried to bridge this gap eitherby fine-tuning or re-training the network at different resolutions. Howeverre-training a network is rarely cheap and not always viable. To this end, wepropose a simple solution to address the train-test distributional shift andenhance the performance of pre-trained models -- which commonly ship as apackage with deep learning platforms \eg, PyTorch. Specifically, we demonstratethat running inference on the center crop of an image is not always the best asimportant discriminatory information may be cropped-off. Instead we propose tocombine results for multiple random crops for a test image. This not onlymatches the train time augmentation but also provides the full coverage of theinput image. We explore combining representation of random crops throughaveraging at different levels \ie, deep feature level, logit level, and softmaxlevel. We demonstrate that, for various families of modern deep networks, suchaveraging results in better validation accuracy compared to using a singlecentral crop per image. The softmax averaging results in the best performancefor various pre-trained networks without requiring any re-training orfine-tuning whatsoever. On modern GPUs with batch processing, the paper'sapproach to inference of pre-trained networks, is essentially free as allimages in a batch can all be processed at once.

Enhanced Performance of Pre-Trained Networks by Matched Augmentation Distributions