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

Even as deep neural networks (DNNs) have achieved remarkable success onvision-related tasks, their performance is brittle to transformations in theinput. Of particular interest are semantic transformations that model changesthat have a basis in the physical world, such as rotations, translations,changes in lighting or camera pose. In this paper, we show how differentiablerendering can be utilized to generate images that are informative, yetrealistic, and which can be used to analyze DNN performance and improve itsrobustness through data augmentation. Given a differentiable renderer and aDNN, we show how to use off-the-shelf attacks from adversarial machine learningto generate semantic counterexamples -- images where semantic features arechanged as to produce misclassifications or misdetections. We validate ourapproach on DNNs for image classification and object detection. Forclassification, we show that semantic counterexamples, when used to augment thedataset, (i) improve generalization performance (ii) enhance robustness tosemantic transformations, and (iii) transfer between models. Additionally, incomparison to sampling-based semantic augmentation, our technique generatesmore informative data in a sample efficient manner.

Analyzing and Improving Neural Networks by Generating Semantic Counterexamples through Differentiable Rendering