5BUILDING DATASETS
The previous chapter had a lot of detailed advice. Now let’s put it all into practice to build the datasets we’ll use throughout the remainder of the book. Some of these datasets are well suited to traditional models, because they consist of feature vectors. Others are better suited to deep learning models that work with multidimensional inputs—in particular, images, or things that can be visualized as images.
We’ll work through acquiring the raw data and preprocessing the data to make it suitable for our tools. We won’t make actual training/validation/test splits until we use these datasets for specific models. It is worth ...
Get Practical Deep Learning now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.
