FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3213702890> ?p ?o ?g. }

• W3213702890 endingPage "3403" @default.
• W3213702890 startingPage "3389" @default.
• W3213702890 abstract "Feature embedding learning and feature interaction modeling are two crucial components of deep models for Click-Through Rate (CTR) prediction in recommender systems. Most existing deep CTR models suffer from the following three problems. First, feature interactions are either manually designed or simply enumerated. However, not all the feature interactions are useful for the prediction task and useless feature interactions may introduce noisy signals thus causing overfitting. Second, all the feature interactions are modeled with an identical interaction function, whereas different interaction functions introduce different inductive biases to better capture various feature interaction patterns. Third, in most existing models, different features share the same embedding size. However, model size can be further optimized without sacrificing performance by differentiating embedding sizes for individual features, as the amount of information contained in each feature varies much. To address the three issues mentioned above, we propose <italic xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>Automatic Interaction Machine (AIM)</i> with three core components, namely, Feature Interaction Search (FIS), Interaction Function Search (IFS) and Embedding Dimension Search (EDS), respectively. To tackle the first problem, FIS component automatically identifies different orders of essential feature interactions with useless ones pruned. Taking care of the second problem, IFS component selects appropriate interaction functions for each individual feature interaction in a learnable way. Moreover, to avoid learning conflict among different interaction functions, IFS proposes function-wise embeddings via performing multiple embeddings for each feature, where each feature embedding corresponds to one possible interaction function. However, utilizing multiple embeddings for each feature may make the model size affordably large if we keep the same embedding size as utilizing shared embedding (i.e., each feature shares the same embedding for different interaction functions). To solve this third problem, EDS automatically selects proper embedding size for each feature. Such a flexible embedding size adaptation is able to reduce the large amount of embedding parameters introduced by function-wise embeddings. Offline experiments on three large-scale datasets (two public benchmarks, one private dataset) validate that AIM can significantly improve various FM-based models. AIM has been deployed in the recommendation service of a mainstream app market, where a three-week online A/B test demonstrated the superiority of AIM, improving DeepFM model by 4.4% in terms of CTR." @default.
• W3213702890 created "2021-11-22" @default.
• W3213702890 creator A5001571390 @default.
• W3213702890 creator A5008281909 @default.
• W3213702890 creator A5037936136 @default.
• W3213702890 creator A5054330014 @default.
• W3213702890 creator A5067483347 @default.
• W3213702890 creator A5077862962 @default.
• W3213702890 creator A5083350101 @default.
• W3213702890 creator A5090720315 @default.
• W3213702890 date "2023-04-01" @default.
• W3213702890 modified "2023-10-15" @default.
• W3213702890 title "AIM: Automatic Interaction Machine for Click-Through Rate Prediction" @default.
• W3213702890 cites W2137983211 @default.
• W3213702890 cites W2295739661 @default.
• W3213702890 cites W2475334473 @default.
• W3213702890 cites W2509235963 @default.
• W3213702890 cites W2512971201 @default.
• W3213702890 cites W2548570154 @default.
• W3213702890 cites W2604662567 @default.
• W3213702890 cites W2723293840 @default.
• W3213702890 cites W2793768763 @default.
• W3213702890 cites W2954817175 @default.
• W3213702890 cites W2963323306 @default.
• W3213702890 cites W2963924287 @default.
• W3213702890 cites W2964052347 @default.
• W3213702890 cites W2964182926 @default.
• W3213702890 cites W2979450518 @default.
• W3213702890 cites W3012731857 @default.
• W3213702890 cites W3034483718 @default.
• W3213702890 cites W3034552531 @default.
• W3213702890 cites W3034906194 @default.
• W3213702890 cites W3081190557 @default.
• W3213702890 cites W3081362488 @default.
• W3213702890 cites W3088777230 @default.
• W3213702890 cites W3093965394 @default.
• W3213702890 cites W3098024612 @default.
• W3213702890 cites W3098723082 @default.
• W3213702890 cites W3101704389 @default.
• W3213702890 cites W3190524507 @default.
• W3213702890 doi "https://doi.org/10.1109/tkde.2021.3134985" @default.
• W3213702890 hasPublicationYear "2023" @default.
• W3213702890 type Work @default.
• W3213702890 sameAs 3213702890 @default.
• W3213702890 citedByCount "0" @default.
• W3213702890 crossrefType "journal-article" @default.
• W3213702890 hasAuthorship W3213702890A5001571390 @default.
• W3213702890 hasAuthorship W3213702890A5008281909 @default.
• W3213702890 hasAuthorship W3213702890A5037936136 @default.
• W3213702890 hasAuthorship W3213702890A5054330014 @default.
• W3213702890 hasAuthorship W3213702890A5067483347 @default.
• W3213702890 hasAuthorship W3213702890A5077862962 @default.
• W3213702890 hasAuthorship W3213702890A5083350101 @default.
• W3213702890 hasAuthorship W3213702890A5090720315 @default.
• W3213702890 hasBestOaLocation W32137028902 @default.
• W3213702890 hasConcept C105795698 @default.
• W3213702890 hasConcept C119857082 @default.
• W3213702890 hasConcept C138885662 @default.
• W3213702890 hasConcept C14036430 @default.
• W3213702890 hasConcept C153180895 @default.
• W3213702890 hasConcept C154945302 @default.
• W3213702890 hasConcept C22019652 @default.
• W3213702890 hasConcept C2776401178 @default.
• W3213702890 hasConcept C33923547 @default.
• W3213702890 hasConcept C38764148 @default.
• W3213702890 hasConcept C41008148 @default.
• W3213702890 hasConcept C41608201 @default.
• W3213702890 hasConcept C41895202 @default.
• W3213702890 hasConcept C50644808 @default.
• W3213702890 hasConcept C78458016 @default.
• W3213702890 hasConcept C80444323 @default.
• W3213702890 hasConcept C83665646 @default.
• W3213702890 hasConcept C86803240 @default.
• W3213702890 hasConceptScore W3213702890C105795698 @default.
• W3213702890 hasConceptScore W3213702890C119857082 @default.
• W3213702890 hasConceptScore W3213702890C138885662 @default.
• W3213702890 hasConceptScore W3213702890C14036430 @default.
• W3213702890 hasConceptScore W3213702890C153180895 @default.
• W3213702890 hasConceptScore W3213702890C154945302 @default.
• W3213702890 hasConceptScore W3213702890C22019652 @default.
• W3213702890 hasConceptScore W3213702890C2776401178 @default.
• W3213702890 hasConceptScore W3213702890C33923547 @default.
• W3213702890 hasConceptScore W3213702890C38764148 @default.
• W3213702890 hasConceptScore W3213702890C41008148 @default.
• W3213702890 hasConceptScore W3213702890C41608201 @default.
• W3213702890 hasConceptScore W3213702890C41895202 @default.
• W3213702890 hasConceptScore W3213702890C50644808 @default.
• W3213702890 hasConceptScore W3213702890C78458016 @default.
• W3213702890 hasConceptScore W3213702890C80444323 @default.
• W3213702890 hasConceptScore W3213702890C83665646 @default.
• W3213702890 hasConceptScore W3213702890C86803240 @default.
• W3213702890 hasFunder F4320321001 @default.
• W3213702890 hasIssue "4" @default.
• W3213702890 hasLocation W32137028901 @default.
• W3213702890 hasLocation W32137028902 @default.
• W3213702890 hasOpenAccess W3213702890 @default.
• W3213702890 hasPrimaryLocation W32137028901 @default.
• W3213702890 hasRelatedWork W2052253960 @default.

• next