Matches in SemOpenAlex for { <https://semopenalex.org/work/W4327989969> ?p ?o ?g. }
Showing items 1 to 69 of
69
with 100 items per page.
- W4327989969 abstract "Image super-resolution (SR) has witnessed extensive neural network designs from CNN to transformer architectures. However, prevailing SR models suffer from prohibitive memory footprint and intensive computations, which limits further deployment on edge devices. This work investigates the potential of network pruning for super-resolution to take advantage of off-the-shelf network designs and reduce the underlying computational overhead. Two main challenges remain in applying pruning methods for SR. First, the widely-used filter pruning technique reflects limited granularity and restricted adaptability to diverse network structures. Second, existing pruning methods generally operate upon a pre-trained network for the sparse structure determination, hard to get rid of dense model training in the traditional SR paradigm. To address these challenges, we adopt unstructured pruning with sparse models directly trained from scratch. Specifically, we propose a novel Iterative Soft Shrinkage-Percentage (ISS-P) method by optimizing the sparse structure of a randomly initialized network at each iteration and tweaking unimportant weights with a small amount proportional to the magnitude scale on-the-fly. We observe that the proposed ISS-P can dynamically learn sparse structures adapting to the optimization process and preserve the sparse model's trainability by yielding a more regularized gradient throughput. Experiments on benchmark datasets demonstrate the effectiveness of the proposed ISS-P over diverse network architectures. Code is available at https://github.com/Jiamian-Wang/Iterative-Soft-Shrinkage-SR" @default.
- W4327989969 created "2023-03-21" @default.
- W4327989969 creator A5000592356 @default.
- W4327989969 creator A5005819096 @default.
- W4327989969 creator A5009988205 @default.
- W4327989969 creator A5021486404 @default.
- W4327989969 creator A5074865219 @default.
- W4327989969 date "2023-03-16" @default.
- W4327989969 modified "2023-09-30" @default.
- W4327989969 title "Iterative Soft Shrinkage Learning for Efficient Image Super-Resolution" @default.
- W4327989969 doi "https://doi.org/10.48550/arxiv.2303.09650" @default.
- W4327989969 hasPublicationYear "2023" @default.
- W4327989969 type Work @default.
- W4327989969 citedByCount "0" @default.
- W4327989969 crossrefType "posted-content" @default.
- W4327989969 hasAuthorship W4327989969A5000592356 @default.
- W4327989969 hasAuthorship W4327989969A5005819096 @default.
- W4327989969 hasAuthorship W4327989969A5009988205 @default.
- W4327989969 hasAuthorship W4327989969A5021486404 @default.
- W4327989969 hasAuthorship W4327989969A5074865219 @default.
- W4327989969 hasBestOaLocation W43279899691 @default.
- W4327989969 hasConcept C108010975 @default.
- W4327989969 hasConcept C111919701 @default.
- W4327989969 hasConcept C113775141 @default.
- W4327989969 hasConcept C11413529 @default.
- W4327989969 hasConcept C115961682 @default.
- W4327989969 hasConcept C11727466 @default.
- W4327989969 hasConcept C119857082 @default.
- W4327989969 hasConcept C13280743 @default.
- W4327989969 hasConcept C154945302 @default.
- W4327989969 hasConcept C185798385 @default.
- W4327989969 hasConcept C205649164 @default.
- W4327989969 hasConcept C41008148 @default.
- W4327989969 hasConcept C50644808 @default.
- W4327989969 hasConcept C6557445 @default.
- W4327989969 hasConcept C74912251 @default.
- W4327989969 hasConcept C86803240 @default.
- W4327989969 hasConceptScore W4327989969C108010975 @default.
- W4327989969 hasConceptScore W4327989969C111919701 @default.
- W4327989969 hasConceptScore W4327989969C113775141 @default.
- W4327989969 hasConceptScore W4327989969C11413529 @default.
- W4327989969 hasConceptScore W4327989969C115961682 @default.
- W4327989969 hasConceptScore W4327989969C11727466 @default.
- W4327989969 hasConceptScore W4327989969C119857082 @default.
- W4327989969 hasConceptScore W4327989969C13280743 @default.
- W4327989969 hasConceptScore W4327989969C154945302 @default.
- W4327989969 hasConceptScore W4327989969C185798385 @default.
- W4327989969 hasConceptScore W4327989969C205649164 @default.
- W4327989969 hasConceptScore W4327989969C41008148 @default.
- W4327989969 hasConceptScore W4327989969C50644808 @default.
- W4327989969 hasConceptScore W4327989969C6557445 @default.
- W4327989969 hasConceptScore W4327989969C74912251 @default.
- W4327989969 hasConceptScore W4327989969C86803240 @default.
- W4327989969 hasLocation W43279899691 @default.
- W4327989969 hasOpenAccess W4327989969 @default.
- W4327989969 hasPrimaryLocation W43279899691 @default.
- W4327989969 hasRelatedWork W1485630101 @default.
- W4327989969 hasRelatedWork W1574999717 @default.
- W4327989969 hasRelatedWork W2135359786 @default.
- W4327989969 hasRelatedWork W3045422318 @default.
- W4327989969 hasRelatedWork W3092479081 @default.
- W4327989969 hasRelatedWork W3162781182 @default.
- W4327989969 hasRelatedWork W4287182679 @default.
- W4327989969 hasRelatedWork W4287704904 @default.
- W4327989969 hasRelatedWork W4319338552 @default.
- W4327989969 hasRelatedWork W4385232115 @default.
- W4327989969 isParatext "false" @default.
- W4327989969 isRetracted "false" @default.
- W4327989969 workType "article" @default.