R Deep Learning Essentials - Second Edition

Data partition between training, test, and validation sets

The previous section highlighted the importance of acquiring some data at an early stage in the project. But if you do not have enough data to train a deep learning model, it is possible to train on other data and apply it to your data. For example, you can use a model trained on ImageNet data for image classification tasks. In this scenario, you need to use the real data that has been collected wisely. This section discusses some good practices on that subject.

If you have ever wondered why big companies such as Google, Apple, Facebook, Amazon, and so on have such a head start in AI, this is the reason why. While they have some of the best AI people in the world working for them, ...