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/

Zendy Plus

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

Zendy Tools

Zendy Open

User response prediction is a crucial component for personalized informationretrieval and filtering scenarios, such as recommender system and web search.The data in user response prediction is mostly in a multi-field categoricalformat and transformed into sparse representations via one-hot encoding. Due tothe sparsity problems in representation and optimization, most research focuseson feature engineering and shallow modeling. Recently, deep neural networkshave attracted research attention on such a problem for their high capacity andend-to-end training scheme. In this paper, we study user response prediction inthe scenario of click prediction. We first analyze a coupled gradient issue inlatent vector-based models and propose kernel product to learn field-awarefeature interactions. Then we discuss an insensitive gradient issue inDNN-based models and propose Product-based Neural Network (PNN) which adopts afeature extractor to explore feature interactions. Generalizing the kernelproduct to a net-in-net architecture, we further propose Product-network InNetwork (PIN) which can generalize previous models. Extensive experiments on 4industrial datasets and 1 contest dataset demonstrate that our modelsconsistently outperform 8 baselines on both AUC and log loss. Besides, PINmakes great CTR improvement (relatively 34.67%) in online A/B test.

Product-based Neural Networks for User Response Prediction over Multi-field Categorical Data