Matches in SemOpenAlex for { <https://semopenalex.org/work/W4385890201> ?p ?o ?g. }
Showing items 1 to 73 of
73
with 100 items per page.
- W4385890201 abstract "Fine-tuning pre-trained neural network models has become a widely adopted approach across various domains. However, it can lead to the distortion of pre-trained feature extractors that already possess strong generalization capabilities. Mitigating feature distortion during adaptation to new target domains is crucial. Recent studies have shown promising results in handling feature distortion by aligning the head layer on in-distribution datasets before performing fine-tuning. Nonetheless, a significant limitation arises from the treatment of batch normalization layers during fine-tuning, leading to suboptimal performance. In this paper, we propose Domain-Aware Fine-Tuning (DAFT), a novel approach that incorporates batch normalization conversion and the integration of linear probing and fine-tuning. Our batch normalization conversion method effectively mitigates feature distortion by reducing modifications to the neural network during fine-tuning. Additionally, we introduce the integration of linear probing and fine-tuning to optimize the head layer with gradual adaptation of the feature extractor. By leveraging batch normalization layers and integrating linear probing and fine-tuning, our DAFT significantly mitigates feature distortion and achieves improved model performance on both in-distribution and out-of-distribution datasets. Extensive experiments demonstrate that our method outperforms other baseline methods, demonstrating its effectiveness in not only improving performance but also mitigating feature distortion." @default.
- W4385890201 created "2023-08-17" @default.
- W4385890201 creator A5035827605 @default.
- W4385890201 creator A5066818142 @default.
- W4385890201 creator A5068058738 @default.
- W4385890201 date "2023-08-15" @default.
- W4385890201 modified "2023-09-27" @default.
- W4385890201 title "Domain-Aware Fine-Tuning: Enhancing Neural Network Adaptability" @default.
- W4385890201 doi "https://doi.org/10.48550/arxiv.2308.07728" @default.
- W4385890201 hasPublicationYear "2023" @default.
- W4385890201 type Work @default.
- W4385890201 citedByCount "0" @default.
- W4385890201 crossrefType "posted-content" @default.
- W4385890201 hasAuthorship W4385890201A5035827605 @default.
- W4385890201 hasAuthorship W4385890201A5066818142 @default.
- W4385890201 hasAuthorship W4385890201A5068058738 @default.
- W4385890201 hasBestOaLocation W43858902011 @default.
- W4385890201 hasConcept C121332964 @default.
- W4385890201 hasConcept C126780896 @default.
- W4385890201 hasConcept C136886441 @default.
- W4385890201 hasConcept C138885662 @default.
- W4385890201 hasConcept C144024400 @default.
- W4385890201 hasConcept C153180895 @default.
- W4385890201 hasConcept C154945302 @default.
- W4385890201 hasConcept C157524613 @default.
- W4385890201 hasConcept C177606310 @default.
- W4385890201 hasConcept C18903297 @default.
- W4385890201 hasConcept C19165224 @default.
- W4385890201 hasConcept C194257627 @default.
- W4385890201 hasConcept C2776257435 @default.
- W4385890201 hasConcept C2776401178 @default.
- W4385890201 hasConcept C31258907 @default.
- W4385890201 hasConcept C41008148 @default.
- W4385890201 hasConcept C41895202 @default.
- W4385890201 hasConcept C50644808 @default.
- W4385890201 hasConcept C62520636 @default.
- W4385890201 hasConcept C86803240 @default.
- W4385890201 hasConceptScore W4385890201C121332964 @default.
- W4385890201 hasConceptScore W4385890201C126780896 @default.
- W4385890201 hasConceptScore W4385890201C136886441 @default.
- W4385890201 hasConceptScore W4385890201C138885662 @default.
- W4385890201 hasConceptScore W4385890201C144024400 @default.
- W4385890201 hasConceptScore W4385890201C153180895 @default.
- W4385890201 hasConceptScore W4385890201C154945302 @default.
- W4385890201 hasConceptScore W4385890201C157524613 @default.
- W4385890201 hasConceptScore W4385890201C177606310 @default.
- W4385890201 hasConceptScore W4385890201C18903297 @default.
- W4385890201 hasConceptScore W4385890201C19165224 @default.
- W4385890201 hasConceptScore W4385890201C194257627 @default.
- W4385890201 hasConceptScore W4385890201C2776257435 @default.
- W4385890201 hasConceptScore W4385890201C2776401178 @default.
- W4385890201 hasConceptScore W4385890201C31258907 @default.
- W4385890201 hasConceptScore W4385890201C41008148 @default.
- W4385890201 hasConceptScore W4385890201C41895202 @default.
- W4385890201 hasConceptScore W4385890201C50644808 @default.
- W4385890201 hasConceptScore W4385890201C62520636 @default.
- W4385890201 hasConceptScore W4385890201C86803240 @default.
- W4385890201 hasLocation W43858902011 @default.
- W4385890201 hasOpenAccess W4385890201 @default.
- W4385890201 hasPrimaryLocation W43858902011 @default.
- W4385890201 hasRelatedWork W1991269640 @default.
- W4385890201 hasRelatedWork W2016839265 @default.
- W4385890201 hasRelatedWork W2136292188 @default.
- W4385890201 hasRelatedWork W2144731606 @default.
- W4385890201 hasRelatedWork W2376729889 @default.
- W4385890201 hasRelatedWork W2382607599 @default.
- W4385890201 hasRelatedWork W2546942002 @default.
- W4385890201 hasRelatedWork W2970216048 @default.
- W4385890201 hasRelatedWork W3035394992 @default.
- W4385890201 hasRelatedWork W2508457823 @default.
- W4385890201 isParatext "false" @default.
- W4385890201 isRetracted "false" @default.
- W4385890201 workType "article" @default.