FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2956587495> ?p ?o ?g. }

• W2956587495 abstract "One of the challenges in display advertising is that the distribution of features and click through rate (CTR) can exhibit large shifts over time due to seasonality, changes to ad campaigns and other factors. The predominant strategy to keep up with these shifts is to train predictive models continuously, on fresh data, in order to prevent them from becoming stale. However, in many ad systems positive labels are only observed after a possibly long and random delay. These delayed labels pose a challenge to data freshness in continuous training: fresh data may not have complete label information at the time they are ingested by the training algorithm. Naive strategies which consider any data point a negative example until a positive label becomes available tend to underestimate CTR, resulting in inferior user experience and suboptimal performance for advertisers. The focus of this paper is to identify the best combination of loss functions and models that enable large-scale learning from a continuous stream of data in the presence of delayed labels. In this work, we compare 5 different loss functions, 3 of them applied to this problem for the first time. We benchmark their performance in offline settings on both public and proprietary datasets in conjunction with shallow and deep model architectures. We also discuss the engineering cost associated with implementing each loss function in a production environment. Finally, we carried out online experiments with the top performing methods, in order to validate their performance in a continuous training scheme. While training on 668 million in-house data points offline, our proposed methods outperform previous state-of-the-art by 3% relative cross entropy (RCE). During online experiments, we observed 55% gain in revenue per thousand requests (RPMq) against naive log loss." @default.
• W2956587495 created "2019-07-23" @default.
• W2956587495 creator A5018244514 @default.
• W2956587495 creator A5029762761 @default.
• W2956587495 creator A5029839691 @default.
• W2956587495 creator A5041064065 @default.
• W2956587495 creator A5052054395 @default.
• W2956587495 creator A5061046370 @default.
• W2956587495 creator A5071533815 @default.
• W2956587495 creator A5074861404 @default.
• W2956587495 date "2019-07-15" @default.
• W2956587495 modified "2023-09-25" @default.
• W2956587495 title "Addressing Delayed Feedback for Continuous Training with Neural Networks in CTR prediction" @default.
• W2956587495 cites W1533861849 @default.
• W2956587495 cites W1825821140 @default.
• W2956587495 cites W1976517433 @default.
• W2956587495 cites W2021866613 @default.
• W2956587495 cites W2049633694 @default.
• W2956587495 cites W2050871273 @default.
• W2956587495 cites W2076618162 @default.
• W2956587495 cites W2101571195 @default.
• W2956587495 cites W2110818436 @default.
• W2956587495 cites W2117864026 @default.
• W2956587495 cites W2122124659 @default.
• W2956587495 cites W2123958887 @default.
• W2956587495 cites W2132442585 @default.
• W2956587495 cites W2133227149 @default.
• W2956587495 cites W2139759436 @default.
• W2956587495 cites W2141570612 @default.
• W2956587495 cites W2146807976 @default.
• W2956587495 cites W2158685856 @default.
• W2956587495 cites W2159566498 @default.
• W2956587495 cites W2475334473 @default.
• W2956587495 cites W2560647685 @default.
• W2956587495 cites W2724479291 @default.
• W2956587495 cites W2785978487 @default.
• W2956587495 cites W2902572901 @default.
• W2956587495 cites W2963335154 @default.
• W2956587495 cites W2963341628 @default.
• W2956587495 cites W2963739929 @default.
• W2956587495 doi "https://doi.org/10.48550/arxiv.1907.06558" @default.
• W2956587495 hasPublicationYear "2019" @default.
• W2956587495 type Work @default.
• W2956587495 sameAs 2956587495 @default.
• W2956587495 citedByCount "3" @default.
• W2956587495 countsByYear W29565874952020 @default.
• W2956587495 countsByYear W29565874952021 @default.
• W2956587495 crossrefType "posted-content" @default.
• W2956587495 hasAuthorship W2956587495A5018244514 @default.
• W2956587495 hasAuthorship W2956587495A5029762761 @default.
• W2956587495 hasAuthorship W2956587495A5029839691 @default.
• W2956587495 hasAuthorship W2956587495A5041064065 @default.
• W2956587495 hasAuthorship W2956587495A5052054395 @default.
• W2956587495 hasAuthorship W2956587495A5061046370 @default.
• W2956587495 hasAuthorship W2956587495A5071533815 @default.
• W2956587495 hasAuthorship W2956587495A5074861404 @default.
• W2956587495 hasBestOaLocation W29565874951 @default.
• W2956587495 hasConcept C108583219 @default.
• W2956587495 hasConcept C110875604 @default.
• W2956587495 hasConcept C115174607 @default.
• W2956587495 hasConcept C119857082 @default.
• W2956587495 hasConcept C120665830 @default.
• W2956587495 hasConcept C121332964 @default.
• W2956587495 hasConcept C13280743 @default.
• W2956587495 hasConcept C134306372 @default.
• W2956587495 hasConcept C136764020 @default.
• W2956587495 hasConcept C139719470 @default.
• W2956587495 hasConcept C14036430 @default.
• W2956587495 hasConcept C153294291 @default.
• W2956587495 hasConcept C154945302 @default.
• W2956587495 hasConcept C162324750 @default.
• W2956587495 hasConcept C185798385 @default.
• W2956587495 hasConcept C192209626 @default.
• W2956587495 hasConcept C205649164 @default.
• W2956587495 hasConcept C23123220 @default.
• W2956587495 hasConcept C2777211547 @default.
• W2956587495 hasConcept C2777999536 @default.
• W2956587495 hasConcept C2778348673 @default.
• W2956587495 hasConcept C33923547 @default.
• W2956587495 hasConcept C41008148 @default.
• W2956587495 hasConcept C50644808 @default.
• W2956587495 hasConcept C512338625 @default.
• W2956587495 hasConcept C51632099 @default.
• W2956587495 hasConcept C77618280 @default.
• W2956587495 hasConcept C78458016 @default.
• W2956587495 hasConcept C86803240 @default.
• W2956587495 hasConceptScore W2956587495C108583219 @default.
• W2956587495 hasConceptScore W2956587495C110875604 @default.
• W2956587495 hasConceptScore W2956587495C115174607 @default.
• W2956587495 hasConceptScore W2956587495C119857082 @default.
• W2956587495 hasConceptScore W2956587495C120665830 @default.
• W2956587495 hasConceptScore W2956587495C121332964 @default.
• W2956587495 hasConceptScore W2956587495C13280743 @default.
• W2956587495 hasConceptScore W2956587495C134306372 @default.
• W2956587495 hasConceptScore W2956587495C136764020 @default.
• W2956587495 hasConceptScore W2956587495C139719470 @default.
• W2956587495 hasConceptScore W2956587495C14036430 @default.
• W2956587495 hasConceptScore W2956587495C153294291 @default.
• W2956587495 hasConceptScore W2956587495C154945302 @default.
• W2956587495 hasConceptScore W2956587495C162324750 @default.

• next