Generating profit with transfer learning
Transfer learning means that we can use a model we designed and trained in another similar case. This will make the model very profitable since we do not have to design a new model and write a new program for every new case. You will thus generate profit for your company or customer by lowering the cost of new implementations of your trained model. Think of a good AI model as a reusable tool when applied to similar cases. This is why concept learning, being more general and abstract, is profitable. That is how we humans adapt.
When it comes to reasoning and thinking in general, we use mental images with some words. Our thoughts contain concepts, on which we build solutions.
The trained model from Chapter 9, Abstract Image Classification with Convolutional Neural Networks (CNNs), can now classify images of a certain type. In this section, the trained model will be loaded and then generalized through transfer learning...