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

Supervisory signals have the potential to make low-dimensional datarepresentations, like those learned by mixture and topic models, moreinterpretable and useful. We propose a framework for training latent variablemodels that explicitly balances two goals: recovery of faithful generativeexplanations of high-dimensional data, and accurate prediction of associatedsemantic labels. Existing approaches fail to achieve these goals due to anincomplete treatment of a fundamental asymmetry: the intended application isalways predicting labels from data, not data from labels. Ourprediction-constrained objective for training generative models coherentlyintegrates loss-based supervisory signals while enabling effectivesemi-supervised learning from partially labeled data. We derive learningalgorithms for semi-supervised mixture and topic models using stochasticgradient descent with automatic differentiation. We demonstrate improvedprediction quality compared to several previous supervised topic models,achieving predictions competitive with high-dimensional logistic regression ontext sentiment analysis and electronic health records tasks whilesimultaneously learning interpretable topics.

Prediction-Constrained Training for Semi-Supervised Mixture and Topic Models