Matches in SemOpenAlex for { <https://semopenalex.org/work/W3007522669> ?p ?o ?g. }
- W3007522669 abstract "Recommender system models often represent various sparse features like users, items, and categorical features via embeddings. A standard approach is to map each unique feature value to an embedding vector. The size of the produced embedding table grows linearly with the size of the vocabulary. Therefore, a large vocabulary inevitably leads to a gigantic embedding table, creating two severe problems: (i) making model serving intractable in resource-constrained environments; (ii) causing overfitting problems. In this paper, we seek to learn highly compact embeddings for large-vocab sparse features in recommender systems (recsys). First, we show that the novel Differentiable Product Quantization (DPQ) approach can generalize to recsys problems. In addition, to better handle the power-law data distribution commonly seen in recsys, we propose a Multi-Granular Quantized Embeddings (MGQE) technique which learns more compact embeddings for infrequent items. We seek to provide a new angle to improve recommendation performance with compact model sizes. Extensive experiments on three recommendation tasks and two datasets show that we can achieve on par or better performance, with only ~20% of the original model size." @default.
- W3007522669 created "2020-03-06" @default.
- W3007522669 creator A5028125399 @default.
- W3007522669 creator A5042916468 @default.
- W3007522669 creator A5050182143 @default.
- W3007522669 creator A5055484401 @default.
- W3007522669 creator A5077716269 @default.
- W3007522669 creator A5079085366 @default.
- W3007522669 creator A5090597599 @default.
- W3007522669 date "2020-02-19" @default.
- W3007522669 modified "2023-10-17" @default.
- W3007522669 title "Learning Multi-granular Quantized Embeddings for Large-Vocab Categorical Features in Recommender Systems" @default.
- W3007522669 cites W1498436455 @default.
- W3007522669 cites W1821462560 @default.
- W3007522669 cites W2041878876 @default.
- W3007522669 cites W2070996757 @default.
- W3007522669 cites W2140679639 @default.
- W3007522669 cites W2153579005 @default.
- W3007522669 cites W2159094788 @default.
- W3007522669 cites W2163605009 @default.
- W3007522669 cites W2194775991 @default.
- W3007522669 cites W2219888463 @default.
- W3007522669 cites W2242818861 @default.
- W3007522669 cites W2300242332 @default.
- W3007522669 cites W2336920772 @default.
- W3007522669 cites W2475334473 @default.
- W3007522669 cites W2583674722 @default.
- W3007522669 cites W2605350416 @default.
- W3007522669 cites W2626376796 @default.
- W3007522669 cites W2734755249 @default.
- W3007522669 cites W2783272285 @default.
- W3007522669 cites W2793768763 @default.
- W3007522669 cites W2808787330 @default.
- W3007522669 cites W2808847742 @default.
- W3007522669 cites W2896457183 @default.
- W3007522669 cites W2912236665 @default.
- W3007522669 cites W2962790997 @default.
- W3007522669 cites W2962943936 @default.
- W3007522669 cites W2963036352 @default.
- W3007522669 cites W2963367478 @default.
- W3007522669 cites W2963403868 @default.
- W3007522669 cites W2963631907 @default.
- W3007522669 cites W2963689957 @default.
- W3007522669 cites W2964182926 @default.
- W3007522669 cites W2964299589 @default.
- W3007522669 cites W2964308564 @default.
- W3007522669 cites W2964316331 @default.
- W3007522669 cites W2964341035 @default.
- W3007522669 cites W2972390490 @default.
- W3007522669 cites W2972801466 @default.
- W3007522669 cites W2973172293 @default.
- W3007522669 cites W2996428491 @default.
- W3007522669 cites W3100848837 @default.
- W3007522669 cites W3116990235 @default.
- W3007522669 doi "https://doi.org/10.48550/arxiv.2002.08530" @default.
- W3007522669 hasPublicationYear "2020" @default.
- W3007522669 type Work @default.
- W3007522669 sameAs 3007522669 @default.
- W3007522669 citedByCount "0" @default.
- W3007522669 crossrefType "posted-content" @default.
- W3007522669 hasAuthorship W3007522669A5028125399 @default.
- W3007522669 hasAuthorship W3007522669A5042916468 @default.
- W3007522669 hasAuthorship W3007522669A5050182143 @default.
- W3007522669 hasAuthorship W3007522669A5055484401 @default.
- W3007522669 hasAuthorship W3007522669A5077716269 @default.
- W3007522669 hasAuthorship W3007522669A5079085366 @default.
- W3007522669 hasAuthorship W3007522669A5090597599 @default.
- W3007522669 hasBestOaLocation W30075226691 @default.
- W3007522669 hasConcept C119857082 @default.
- W3007522669 hasConcept C138885662 @default.
- W3007522669 hasConcept C154945302 @default.
- W3007522669 hasConcept C22019652 @default.
- W3007522669 hasConcept C2776401178 @default.
- W3007522669 hasConcept C41008148 @default.
- W3007522669 hasConcept C41608201 @default.
- W3007522669 hasConcept C41895202 @default.
- W3007522669 hasConcept C50644808 @default.
- W3007522669 hasConcept C5274069 @default.
- W3007522669 hasConcept C557471498 @default.
- W3007522669 hasConcept C80444323 @default.
- W3007522669 hasConceptScore W3007522669C119857082 @default.
- W3007522669 hasConceptScore W3007522669C138885662 @default.
- W3007522669 hasConceptScore W3007522669C154945302 @default.
- W3007522669 hasConceptScore W3007522669C22019652 @default.
- W3007522669 hasConceptScore W3007522669C2776401178 @default.
- W3007522669 hasConceptScore W3007522669C41008148 @default.
- W3007522669 hasConceptScore W3007522669C41608201 @default.
- W3007522669 hasConceptScore W3007522669C41895202 @default.
- W3007522669 hasConceptScore W3007522669C50644808 @default.
- W3007522669 hasConceptScore W3007522669C5274069 @default.
- W3007522669 hasConceptScore W3007522669C557471498 @default.
- W3007522669 hasConceptScore W3007522669C80444323 @default.
- W3007522669 hasLocation W30075226691 @default.
- W3007522669 hasOpenAccess W3007522669 @default.
- W3007522669 hasPrimaryLocation W30075226691 @default.
- W3007522669 hasRelatedWork W1996541855 @default.
- W3007522669 hasRelatedWork W2985459377 @default.
- W3007522669 hasRelatedWork W2989932438 @default.
- W3007522669 hasRelatedWork W3011996705 @default.
- W3007522669 hasRelatedWork W3099765033 @default.