高级检索
刘真, 颜菁, 吴兆国, 林菲, 吴向阳. 诊断和提高迁移学习模型鲁棒性的可视分析方法[J]. 计算机辅助设计与图形学学报. DOI: 10.3724/SP.J.1089.null.2023-00585
引用本文: 刘真, 颜菁, 吴兆国, 林菲, 吴向阳. 诊断和提高迁移学习模型鲁棒性的可视分析方法[J]. 计算机辅助设计与图形学学报. DOI: 10.3724/SP.J.1089.null.2023-00585
Zhen Liu, Jing Yan, Zhaoguo Wu, Fei Lin, Xiangyang Wu. A Visual Framework for Diagnosing and Mitigating the Robustness of Transfer Learning Models[J]. Journal of Computer-Aided Design & Computer Graphics. DOI: 10.3724/SP.J.1089.null.2023-00585
Citation: Zhen Liu, Jing Yan, Zhaoguo Wu, Fei Lin, Xiangyang Wu. A Visual Framework for Diagnosing and Mitigating the Robustness of Transfer Learning Models[J]. Journal of Computer-Aided Design & Computer Graphics. DOI: 10.3724/SP.J.1089.null.2023-00585

诊断和提高迁移学习模型鲁棒性的可视分析方法

A Visual Framework for Diagnosing and Mitigating the Robustness of Transfer Learning Models

  • 摘要:       迁移学习作为深度学习社区里一种流行的模型重用技术, 它使开发人员能够根据复杂的预训练模型 (教师模型) 构建符合目标任务的自定义模型 (学生模型). 然而, 与传统软件重用中的缺陷继承一样, 迁移学习中的学生模型也可能继承教师模型的缺陷, 模型的这种缺陷体现为模型的对抗鲁棒性. 在深度学习领域中, 通常运用缺陷缓解或学生模型和教师模型联合训练的方法, 达到减少继承教师模型的缺陷并提高模型对抗鲁棒性的目的. 这些方法普遍存在效率偏低且不够通用的问题, 因此本文提出了一个用于探索迁移学习过程中模型对抗鲁棒性变化的可视分析框架. 该框架中加入相关模型切片的方法以缓解模型缺陷继承, 从模型整体性能、数据实例、实例特征和模型局部结构等多个维度上查看和诊断教师模型和学生模型的性能变化和底层行为机制. 实验表明该框架可以帮助用户分析迁移学习中模型的鲁棒性、模型继承的缺陷知识和模型缺陷改善后的性能变化.

     

    Abstract:     Transfer learning, a popular model reuse technique in the deep learning community, enables developers to build custom models (student models) that fit the target task based on complex pre-trained models (teacher models). However, just like the defect inheritance in traditional software reuse, the student model in transfer learning may also inherit the defect of the teacher model, which is reflected in the adversarial robustness of the model. In the field of deep learning, the method of defect mitigation or joint training of a student model and a teacher model is usually used to reduce the defects of the inherited teacher model and improve the robustness of the model. These methods are generally inefficient and not general, so this paper proposes a visual analysis framework to explore changes in the adversarial robustness of model. In this framework, the method of model slicing is added to alleviate model defect inheritance, and the performance changes and underlying behavior mechanism of the teacher model and the student model are checked and diagnosed from multiple dimensions such as overall model performance, data instance, instance feature and local structure of model. Experiments show that this framework can help users analyze the robustness of the model in transfer learning, the inherited defect knowledge of the model and the performance change after the im-provement of the model defect.

     

/

返回文章
返回