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

Programs offer compactness and structure that makes them an attractiverepresentation for visual data. We explore how code rewriting can be used toimprove systems for inferring programs from visual data. We first proposeSparse Intermittent Rewrite Injection (SIRI), a framework for unsupervisedbootstrapped learning. SIRI sparsely applies code rewrite operations over adataset of training programs, injecting the improved programs back into thetraining set. We design a family of rewriters for visual programming domains:parameter optimization, code pruning, and code grafting. For three shapeprogramming languages in 2D and 3D, we show that using SIRI with our family ofrewriters improves performance: better reconstructions and faster convergencerates, compared with bootstrapped learning methods that do not use rewriters oruse them naively. Finally, we demonstrate that our family of rewriters can beeffectively used at test time to improve the output of SIRI predictions. For 2Dand 3D CSG, we outperform or match the reconstruction performance of recentdomain-specific neural architectures, while producing more parsimoniousprograms that use significantly fewer primitives.

Improving Unsupervised Visual Program Inference with Code Rewriting Families