arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2601.11235cs.CV
收藏

生物启发的细调技术用于图像分类中的选择性迁移学习

Bio-inspired fine-tuning for selective transfer learning in image classification

Institutes of Innovation for Future Society(未来创新研究所);Nagoya University(名古屋大学)

Ana Davila, Jacinto Colan, Yasuhisa Hasegawa

AI总结:

BioTune通过生物启发的进化优化技术,提升图像分类中选择性迁移学习的精度与效率,适用于多种数据特征和分布变化。

摘要:

深度学习在图像分析的多样化领域中显著提高了性能,但通常依赖于大规模标注数据集才能取得成功。迁移学习通过利用预训练模型来处理新任务,以有限的标注数据应对这一挑战。然而,源域和目标域之间的差异可能会阻碍有效的迁移学习。我们引入了BioTune,一种新的自适应细调技术,利用进化优化。BioTune通过最优选择冻结哪些层以及调整未冻结层的学习率来增强迁移学习。通过在九个图像分类数据集上的广泛评估,涵盖自然和专业领域如医学影像,BioTune在精度和效率上均优于最先进的细调方法,包括AutoRGN和LoRA,突显了其对各种数据特征和分布变化的适应性。此外,BioTune在四种不同的CNN架构上始终取得顶尖表现,凸显了其灵活性。消融研究提供了关于BioTune关键组件对整体性能影响的有价值见解。源代码可在https://github.com/davilac/BioTune上获得。

英文摘要:

Deep learning has significantly advanced image analysis across diverse domains but often depends on large, annotated datasets for success. Transfer learning addresses this challenge by utilizing pre-trained models to tackle new tasks with limited labeled data. However, discrepancies between source and target domains can hinder effective transfer learning. We introduce BioTune, a novel adaptive fine-tuning technique utilizing evolutionary optimization. BioTune enhances transfer learning by optimally choosing which layers to freeze and adjusting learning rates for unfrozen layers. Through extensive evaluation on nine image classification datasets, spanning natural and specialized domains such as medical imaging, BioTune demonstrates superior accuracy and efficiency over state-of-the-art fine-tuning methods, including AutoRGN and LoRA, highlighting its adaptability to various data characteristics and distribution changes. Additionally, BioTune consistently achieves top performance across four different CNN architectures, underscoring its flexibility. Ablation studies provide valuable insights into the impact of BioTune's key components on overall performance. The source code is available at https://github.com/davilac/BioTune.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑