Matches in SemOpenAlex for { <https://semopenalex.org/work/W2972269283> ?p ?o ?g. }
Showing items 1 to 84 of
84
with 100 items per page.
- W2972269283 abstract "Modern deep learning-based recommendation systems exploit hundreds to thousands of different categorical features, each with millions of different categories ranging from clicks to posts. To respect the natural diversity within the categorical data, embeddings map each category to a unique dense representation within an embedded space. Since each categorical feature could take on as many as tens of millions of different possible categories, the embedding tables form the primary memory bottleneck during both training and inference. We propose a novel approach for reducing the embedding size in an end-to-end fashion by exploiting complementary partitions of the category set to produce a unique embedding vector for each category without explicit definition. By storing multiple smaller embedding tables based on each complementary partition and combining embeddings from each table, we define a unique embedding for each category at smaller cost. This approach may be interpreted as using a specific fixed codebook to ensure uniqueness of each category's representation. Our experimental results demonstrate the effectiveness of our approach over the hashing trick for reducing the size of the embedding tables in terms of model loss and accuracy, while retaining a similar reduction in the number of parameters." @default.
- W2972269283 created "2019-09-12" @default.
- W2972269283 creator A5033659558 @default.
- W2972269283 creator A5038603364 @default.
- W2972269283 creator A5057574221 @default.
- W2972269283 creator A5066366329 @default.
- W2972269283 date "2020-08-20" @default.
- W2972269283 modified "2023-10-02" @default.
- W2972269283 title "Compositional Embeddings Using Complementary Partitions for Memory-Efficient Recommendation Systems" @default.
- W2972269283 cites W2094286023 @default.
- W2972269283 cites W2295739661 @default.
- W2972269283 cites W2475334473 @default.
- W2972269283 cites W2605350416 @default.
- W2972269283 cites W2723293840 @default.
- W2972269283 cites W2783666221 @default.
- W2972269283 cites W2793768763 @default.
- W2972269283 cites W2963981376 @default.
- W2972269283 cites W2964182926 @default.
- W2972269283 cites W2998508934 @default.
- W2972269283 doi "https://doi.org/10.1145/3394486.3403059" @default.
- W2972269283 hasPublicationYear "2020" @default.
- W2972269283 type Work @default.
- W2972269283 sameAs 2972269283 @default.
- W2972269283 citedByCount "29" @default.
- W2972269283 countsByYear W29722692832020 @default.
- W2972269283 countsByYear W29722692832021 @default.
- W2972269283 countsByYear W29722692832022 @default.
- W2972269283 countsByYear W29722692832023 @default.
- W2972269283 crossrefType "proceedings-article" @default.
- W2972269283 hasAuthorship W2972269283A5033659558 @default.
- W2972269283 hasAuthorship W2972269283A5038603364 @default.
- W2972269283 hasAuthorship W2972269283A5057574221 @default.
- W2972269283 hasAuthorship W2972269283A5066366329 @default.
- W2972269283 hasBestOaLocation W29722692832 @default.
- W2972269283 hasConcept C114614502 @default.
- W2972269283 hasConcept C119857082 @default.
- W2972269283 hasConcept C127759330 @default.
- W2972269283 hasConcept C133667856 @default.
- W2972269283 hasConcept C138111711 @default.
- W2972269283 hasConcept C153180895 @default.
- W2972269283 hasConcept C154945302 @default.
- W2972269283 hasConcept C33923547 @default.
- W2972269283 hasConcept C38652104 @default.
- W2972269283 hasConcept C41008148 @default.
- W2972269283 hasConcept C41608201 @default.
- W2972269283 hasConcept C42812 @default.
- W2972269283 hasConcept C5274069 @default.
- W2972269283 hasConcept C67388219 @default.
- W2972269283 hasConcept C80444323 @default.
- W2972269283 hasConcept C99138194 @default.
- W2972269283 hasConceptScore W2972269283C114614502 @default.
- W2972269283 hasConceptScore W2972269283C119857082 @default.
- W2972269283 hasConceptScore W2972269283C127759330 @default.
- W2972269283 hasConceptScore W2972269283C133667856 @default.
- W2972269283 hasConceptScore W2972269283C138111711 @default.
- W2972269283 hasConceptScore W2972269283C153180895 @default.
- W2972269283 hasConceptScore W2972269283C154945302 @default.
- W2972269283 hasConceptScore W2972269283C33923547 @default.
- W2972269283 hasConceptScore W2972269283C38652104 @default.
- W2972269283 hasConceptScore W2972269283C41008148 @default.
- W2972269283 hasConceptScore W2972269283C41608201 @default.
- W2972269283 hasConceptScore W2972269283C42812 @default.
- W2972269283 hasConceptScore W2972269283C5274069 @default.
- W2972269283 hasConceptScore W2972269283C67388219 @default.
- W2972269283 hasConceptScore W2972269283C80444323 @default.
- W2972269283 hasConceptScore W2972269283C99138194 @default.
- W2972269283 hasLocation W29722692831 @default.
- W2972269283 hasLocation W29722692832 @default.
- W2972269283 hasOpenAccess W2972269283 @default.
- W2972269283 hasPrimaryLocation W29722692831 @default.
- W2972269283 hasRelatedWork W1972853352 @default.
- W2972269283 hasRelatedWork W2573350151 @default.
- W2972269283 hasRelatedWork W2950026462 @default.
- W2972269283 hasRelatedWork W2972269283 @default.
- W2972269283 hasRelatedWork W3010151463 @default.
- W2972269283 hasRelatedWork W3128011138 @default.
- W2972269283 hasRelatedWork W3176468443 @default.
- W2972269283 hasRelatedWork W3196655819 @default.
- W2972269283 hasRelatedWork W3199771031 @default.
- W2972269283 hasRelatedWork W4282940688 @default.
- W2972269283 isParatext "false" @default.
- W2972269283 isRetracted "false" @default.
- W2972269283 magId "2972269283" @default.
- W2972269283 workType "article" @default.