Transfer Learning

Transfer learning reuses a network trained on one large dataset as a starting point for a different task. Here the convolutional layers of VGG16, trained on the 1.3 million images of ImageNet, are frozen and used as a feature extractor for CIFAR-10, and only a small classifier on top is trained.

The benefit is largest when labelled examples are few, which is the usual situation for rare events in heliophysics. The notebooks therefore train on 1,000, 5,000, and 45,000 images and compare the transfer model with a small network trained from scratch on the same images.

• PyTorch notebook

• Keras 3 notebook