[Linear Probing | Linear Detection] Deep Learning Linear Layer
1. Function
The self-supervised model evaluation method
is a method to test the performance of the pre-trained model, also known as linear probing evaluation
2. Principle
After training, to evaluate the quality of the model, by replacing the last layer with a linear layer.
The characteristics of the representation layer of the pre-training model are fixed, and the parameters have not changed after curing. Only supervised data is used to train the classifier (usually Softmax classifier or SVM classifier, etc.).
Only training this linear layer is a linear probe.
3. Provenance
He Yuming MAE
https://zhuanlan.zhihu.com/p/432614068
4. Reference
https://blog.csdn.net/LoseInVain/article/details/103870157
https://blog.csdn.net/qq_23981335/article/details/122576120