Structure of DANN model. The legend explains the activation function, the network connection mode, and the direction of information propagation. (IMAGE)
Caption
Compared with the transfer strategy based on fine-tuning, the domain adversarial neural network (DANN) fine-tuning strategy based on adversarial thinking can improve the Z-class prediction accuracy of the coated material in a completely unlabeled scenario, which greatly enhances the feasibility of applying transfer learning methods in muon Z-class identification
Credit
Liang-Wen Chen
Usage Restrictions
Credit must be given to the creator. Only noncommercial uses of the work are permitted.
License
CC BY-NC