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

Modern semantic parsers suffer from two principal limitations. First,training requires expensive collection of utterance-program pairs. Second,semantic parsers fail to generalize at test time to new compositions/structuresthat have not been observed during training. Recent research has shown thatautomatic generation of synthetic utterance-program pairs can alleviate thefirst problem, but its potential for the second has thus far beenunder-explored. In this work, we investigate automatic generation of syntheticutterance-program pairs for improving compositional generalization in semanticparsing. Given a small training set of annotated examples and an "infinite"pool of synthetic examples, we select a subset of synthetic examples that arestructurally-diverse and use them to improve compositional generalization. Weevaluate our approach on a new split of the schema2QA dataset, and show that itleads to dramatic improvements in compositional generalization as well asmoderate improvements in the traditional i.i.d setup. Moreover,structurally-diverse sampling achieves these improvements with as few as 5Kexamples, compared to 1M examples when sampling uniformly at random -- a 200ximprovement in data efficiency.

Finding needles in a haystack: Sampling Structurally-diverse Training Sets from Synthetic Data for Compositional Generalization