10 sett 2019 anni - 针对推导迁移学习的正则化
使用论文里提出的 L2-SP 代替 L2 惩罚
A baseline regularization scheme for transfer learning with convolutional neural networks
Xuhong Li, Yves Grandvalet, Franck Davoine
2019.9
Descrizione:
文中指出 fine-tuning 可能会导致本来已学到的知识丢失。
Aggiunto al nastro di tempo:
Data:
~ 4 years and 8 months ago