Keras 2.x Projects
上QQ阅读APP看书,第一时间看更新

Unsupervised learning

The aim of unsupervised learning is to extract information from databases automatically . This process occurs without prior knowledge of the contents to be analyzed. Unlike supervised learning, there's no information on membership classes of the examples or generally on the output corresponding to a certain input. The goal is to get a model that's able to discover interesting properties; groups with similar characteristics (clustering), for instance. Search engines are an example of an application of these algorithms. Given one or more keywords, they're able to create a list of links related to our search.

The validity of these algorithms depends on the usefulness of the information they can extract from the databases. These algorithms work by comparing data and looking for similarities or differences. Available data concerns only the set of features that describe each example. They show great efficiency with elements of numeric type, but are much less accurate with nonnumeric data. Generally, they work properly in the presence of data that contains an order or a clear grouping and is clearly identifiable.

The following diagram shows a generic unsupervised learning workflow:

In the previous diagram, a series of images of fruits characterized by different shapes and colors are supplied as input. Nothing else is provided as input. The neural network will be able to group objects (images) on the basis of some common properties, which may be the color (yellow and orange) or the form (elliptical and rounded).