FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4386075538> ?p ?o ?g. }

• W4386075538 abstract "We revisit and advance visual prompting (VP), an input prompting technique for vision tasks. VP can reprogram a fixed, pre-trained source model to accomplish downstream tasks in the target domain by simply incorporating universal prompts (in terms of input perturbation patterns) into downstream data points. Yet, it remains elusive why VP stays effective even given a ruleless label mapping (LM) between the source classes and the target classes. Inspired by the above, we ask: How is LM interrelated with VP? And how to exploit such a relationship to improve its accuracy on target tasks? We peer into the influence of LM on VP and provide an affirmative answer that a better ‘quality’ of LM (assessed by mapping precision and explanation) can consistently improve the effectiveness of VP. This is in contrast to the prior art where the factor of LM was missing. To optimize LM, we propose a new VP framework, termed ILM-VP (iterative label mapping-based visual prompting), which automatically re-maps the source labels to the target labels and progressively improves the target task accuracy of VP. Further, when using a contrastive language-image pretrained (CLIP) model for VP, we propose to integrate an LM process to assist the text prompt selection of CLIP and to improve the target task accuracy. Extensive experiments demonstrate that our proposal significantly outperforms state-of-the-art VP methods. As highlighted below, we show that when reprogramming an ImageNet-pretrained ResNet-18 to 13 target tasks, ILM-VP outperforms baselines by a substantial margin, e.g., 7.9% and 6.7% accuracy improvements in transfer learning to the target Flowers102 and CIFAR100 datasets. Besides, our proposal on CLIP-based VP provides 13.7% and 7.1% accuracy improvements on Flowers102 and DTD respectively. Code is available at https://github.com/OPTML-Group/ILM-VP." @default.
• W4386075538 created "2023-08-23" @default.
• W4386075538 creator A5002976916 @default.
• W4386075538 creator A5008614366 @default.
• W4386075538 creator A5032224340 @default.
• W4386075538 creator A5050344371 @default.
• W4386075538 creator A5059826739 @default.
• W4386075538 date "2023-06-01" @default.
• W4386075538 modified "2023-09-27" @default.
• W4386075538 title "Understanding and Improving Visual Prompting: A Label-Mapping Perspective" @default.
• W4386075538 cites W1977295328 @default.
• W4386075538 cites W2002427601 @default.
• W4386075538 cites W2017814585 @default.
• W4386075538 cites W2047643928 @default.
• W4386075538 cites W2108598243 @default.
• W4386075538 cites W2115403315 @default.
• W4386075538 cites W2115575686 @default.
• W4386075538 cites W2122922389 @default.
• W4386075538 cites W2125792038 @default.
• W4386075538 cites W2138011018 @default.
• W4386075538 cites W2165698076 @default.
• W4386075538 cites W2194775991 @default.
• W4386075538 cites W2533598788 @default.
• W4386075538 cites W2549139847 @default.
• W4386075538 cites W2963168418 @default.
• W4386075538 cites W2964194231 @default.
• W4386075538 cites W3171007011 @default.
• W4386075538 cites W3174770825 @default.
• W4386075538 cites W3210278576 @default.
• W4386075538 cites W4205991051 @default.
• W4386075538 cites W4213328902 @default.
• W4386075538 cites W4386065871 @default.
• W4386075538 doi "https://doi.org/10.1109/cvpr52729.2023.01834" @default.
• W4386075538 hasPublicationYear "2023" @default.
• W4386075538 type Work @default.
• W4386075538 citedByCount "0" @default.
• W4386075538 crossrefType "proceedings-article" @default.
• W4386075538 hasAuthorship W4386075538A5002976916 @default.
• W4386075538 hasAuthorship W4386075538A5008614366 @default.
• W4386075538 hasAuthorship W4386075538A5032224340 @default.
• W4386075538 hasAuthorship W4386075538A5050344371 @default.
• W4386075538 hasAuthorship W4386075538A5059826739 @default.
• W4386075538 hasConcept C119857082 @default.
• W4386075538 hasConcept C12713177 @default.
• W4386075538 hasConcept C153180895 @default.
• W4386075538 hasConcept C154945302 @default.
• W4386075538 hasConcept C162324750 @default.
• W4386075538 hasConcept C165696696 @default.
• W4386075538 hasConcept C187736073 @default.
• W4386075538 hasConcept C2780451532 @default.
• W4386075538 hasConcept C38652104 @default.
• W4386075538 hasConcept C41008148 @default.
• W4386075538 hasConcept C774472 @default.
• W4386075538 hasConceptScore W4386075538C119857082 @default.
• W4386075538 hasConceptScore W4386075538C12713177 @default.
• W4386075538 hasConceptScore W4386075538C153180895 @default.
• W4386075538 hasConceptScore W4386075538C154945302 @default.
• W4386075538 hasConceptScore W4386075538C162324750 @default.
• W4386075538 hasConceptScore W4386075538C165696696 @default.
• W4386075538 hasConceptScore W4386075538C187736073 @default.
• W4386075538 hasConceptScore W4386075538C2780451532 @default.
• W4386075538 hasConceptScore W4386075538C38652104 @default.
• W4386075538 hasConceptScore W4386075538C41008148 @default.
• W4386075538 hasConceptScore W4386075538C774472 @default.
• W4386075538 hasFunder F4320306076 @default.
• W4386075538 hasLocation W43860755381 @default.
• W4386075538 hasOpenAccess W4386075538 @default.
• W4386075538 hasPrimaryLocation W43860755381 @default.
• W4386075538 hasRelatedWork W1607315280 @default.
• W4386075538 hasRelatedWork W2331043530 @default.
• W4386075538 hasRelatedWork W2374725260 @default.
• W4386075538 hasRelatedWork W2393933887 @default.
• W4386075538 hasRelatedWork W2961085424 @default.
• W4386075538 hasRelatedWork W2983785000 @default.
• W4386075538 hasRelatedWork W2997512100 @default.
• W4386075538 hasRelatedWork W4306674287 @default.
• W4386075538 hasRelatedWork W4386140649 @default.
• W4386075538 hasRelatedWork W4224009465 @default.
• W4386075538 isParatext "false" @default.
• W4386075538 isRetracted "false" @default.
• W4386075538 workType "article" @default.