Matches in SemOpenAlex for { <https://semopenalex.org/work/W3210117396> ?p ?o ?g. }
Showing items 1 to 95 of
95
with 100 items per page.
- W3210117396 abstract "The discovery and prediction of block access patterns in hybrid storage systems is of crucial importance for effective tier management. Existing methods are usually based on heuristics and unable to handle complex patterns. This work newly introduces transformer to block access pattern prediction. We remark that block accesses in the tier management systems are aggregated temporally and spatially as multivariate time series of block access frequency, so the runtime requirements are relaxed, making complex models applicable for the deployment. Moreover, enormous and rarely accessed blocks in storage systems and the structure of traditional transformer models would result in millions of redundant parameters and make them impractical to be deployed. We incorporate Tensor-Train Decomposition (TTD) with transformer and propose the Compressed Full Tenor Transformer (CFTT), in which all linear layers in the vanilla transformer are replaced with tensor-train layers. Weights of input and output layers are shared to further reduce parameters and reuse knowledge implicitly. CFTT can significantly reduce the model size and computation cost, which is critical to save storage space and inference time. Extensive experiments are conducted on synthetic and real-world datasets. The results demonstrate that transformers achieve state-of-the-art performance stably in terms of top-k hit rates. Moreover, the proposed CFTT compresses transformers 16× to 461× and speeds up inference 5× without sacrificing performance on the whole, which facilitates its applications in tier management in hybrid storage systems." @default.
- W3210117396 created "2021-11-08" @default.
- W3210117396 creator A5000497132 @default.
- W3210117396 creator A5010738468 @default.
- W3210117396 creator A5019084435 @default.
- W3210117396 creator A5024427207 @default.
- W3210117396 creator A5038698223 @default.
- W3210117396 creator A5062026103 @default.
- W3210117396 creator A5068714594 @default.
- W3210117396 creator A5078949174 @default.
- W3210117396 creator A5088383943 @default.
- W3210117396 date "2021-10-26" @default.
- W3210117396 modified "2023-10-16" @default.
- W3210117396 title "Block Access Pattern Discovery via Compressed Full Tensor Transformer" @default.
- W3210117396 cites W1963830244 @default.
- W3210117396 cites W1993482030 @default.
- W3210117396 cites W2130125728 @default.
- W3210117396 cites W2201220957 @default.
- W3210117396 cites W2263002732 @default.
- W3210117396 cites W2344094242 @default.
- W3210117396 cites W2604847698 @default.
- W3210117396 cites W2885965959 @default.
- W3210117396 cites W2891008051 @default.
- W3210117396 cites W2903739847 @default.
- W3210117396 cites W2963394925 @default.
- W3210117396 cites W2963979055 @default.
- W3210117396 cites W2982027745 @default.
- W3210117396 cites W2997857369 @default.
- W3210117396 cites W3025987117 @default.
- W3210117396 doi "https://doi.org/10.1145/3459637.3482323" @default.
- W3210117396 hasPublicationYear "2021" @default.
- W3210117396 type Work @default.
- W3210117396 sameAs 3210117396 @default.
- W3210117396 citedByCount "2" @default.
- W3210117396 countsByYear W32101173962023 @default.
- W3210117396 crossrefType "proceedings-article" @default.
- W3210117396 hasAuthorship W3210117396A5000497132 @default.
- W3210117396 hasAuthorship W3210117396A5010738468 @default.
- W3210117396 hasAuthorship W3210117396A5019084435 @default.
- W3210117396 hasAuthorship W3210117396A5024427207 @default.
- W3210117396 hasAuthorship W3210117396A5038698223 @default.
- W3210117396 hasAuthorship W3210117396A5062026103 @default.
- W3210117396 hasAuthorship W3210117396A5068714594 @default.
- W3210117396 hasAuthorship W3210117396A5078949174 @default.
- W3210117396 hasAuthorship W3210117396A5088383943 @default.
- W3210117396 hasConcept C111919701 @default.
- W3210117396 hasConcept C113775141 @default.
- W3210117396 hasConcept C11413529 @default.
- W3210117396 hasConcept C119599485 @default.
- W3210117396 hasConcept C120314980 @default.
- W3210117396 hasConcept C127413603 @default.
- W3210117396 hasConcept C127705205 @default.
- W3210117396 hasConcept C154945302 @default.
- W3210117396 hasConcept C165801399 @default.
- W3210117396 hasConcept C206588197 @default.
- W3210117396 hasConcept C2776214188 @default.
- W3210117396 hasConcept C41008148 @default.
- W3210117396 hasConcept C45374587 @default.
- W3210117396 hasConcept C548081761 @default.
- W3210117396 hasConcept C66322947 @default.
- W3210117396 hasConcept C79403827 @default.
- W3210117396 hasConceptScore W3210117396C111919701 @default.
- W3210117396 hasConceptScore W3210117396C113775141 @default.
- W3210117396 hasConceptScore W3210117396C11413529 @default.
- W3210117396 hasConceptScore W3210117396C119599485 @default.
- W3210117396 hasConceptScore W3210117396C120314980 @default.
- W3210117396 hasConceptScore W3210117396C127413603 @default.
- W3210117396 hasConceptScore W3210117396C127705205 @default.
- W3210117396 hasConceptScore W3210117396C154945302 @default.
- W3210117396 hasConceptScore W3210117396C165801399 @default.
- W3210117396 hasConceptScore W3210117396C206588197 @default.
- W3210117396 hasConceptScore W3210117396C2776214188 @default.
- W3210117396 hasConceptScore W3210117396C41008148 @default.
- W3210117396 hasConceptScore W3210117396C45374587 @default.
- W3210117396 hasConceptScore W3210117396C548081761 @default.
- W3210117396 hasConceptScore W3210117396C66322947 @default.
- W3210117396 hasConceptScore W3210117396C79403827 @default.
- W3210117396 hasFunder F4320335777 @default.
- W3210117396 hasLocation W32101173961 @default.
- W3210117396 hasOpenAccess W3210117396 @default.
- W3210117396 hasPrimaryLocation W32101173961 @default.
- W3210117396 hasRelatedWork W1853785581 @default.
- W3210117396 hasRelatedWork W2910518169 @default.
- W3210117396 hasRelatedWork W3130954105 @default.
- W3210117396 hasRelatedWork W3146091044 @default.
- W3210117396 hasRelatedWork W3203166921 @default.
- W3210117396 hasRelatedWork W4214588794 @default.
- W3210117396 hasRelatedWork W4283714288 @default.
- W3210117396 hasRelatedWork W4287241953 @default.
- W3210117396 hasRelatedWork W4361193654 @default.
- W3210117396 hasRelatedWork W4385574170 @default.
- W3210117396 isParatext "false" @default.
- W3210117396 isRetracted "false" @default.
- W3210117396 magId "3210117396" @default.
- W3210117396 workType "article" @default.