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

Program synthesis is the task of automatically generating a programconsistent with a specification. Recent years have seen proposal of a number ofneural approaches for program synthesis, many of which adopt a sequencegeneration paradigm similar to neural machine translation, in whichsequence-to-sequence models are trained to maximize the likelihood of knownreference programs. While achieving impressive results, this strategy has twokey limitations. First, it ignores Program Aliasing: the fact that manydifferent programs may satisfy a given specification (especially withincomplete specifications such as a few input-output examples). By maximizingthe likelihood of only a single reference program, it penalizes manysemantically correct programs, which can adversely affect the synthesizerperformance. Second, this strategy overlooks the fact that programs have astrict syntax that can be efficiently checked. To address the first limitation,we perform reinforcement learning on top of a supervised model with anobjective that explicitly maximizes the likelihood of generating semanticallycorrect programs. For addressing the second limitation, we introduce a trainingprocedure that directly maximizes the probability of generating syntacticallycorrect programs that fulfill the specification. We show that our contributionslead to improved accuracy of the models, especially in cases where the trainingdata is limited.

Leveraging Grammar and Reinforcement Learning for Neural Program Synthesis