Hands-On One-shot Learning with Python

Transfer learning

Transfer learning refers to the technique of using knowledge gleaned from solving one problem and using that to solve a different problem. The following is an illustration of a simplistic view of the transfer learning approach:

In other words, a neural network model trained on one dataset can be used for other datasets by fine-tuning the former network, just like how we can use Siamese networks trained on different domain datasets (such as the MNIST dataset) to extract better features for signature matching, handwriting matching, and so on. Transfer learning has attracted a lot of attention in the field of deep learning and ...