Fig. 2From: Reducing the number of unnecessary biopsies for mammographic BI-RADS 4 lesions through a deep transfer learning methodSchematic diagram of fine-tuning strategies for Inception V3. There were eleven fine-tuning strategies in total. Note: Trainable params, the number of trainable parameters. Trainable layer: activated layers of the neural network. Non-trainable layer: frozen layersBack to article page