Unsupervised learning

This product was added to computer science because of the content, defects on the quality assurance side of the editor. This is done to bring the quality of the articles from the computer science subject area to an acceptable level. Help us to eliminate the substantive shortcomings of this article and take part you in the discussion! ( )

Unsupervised learning (English unsupervised learning) refers to machine learning without pre- known target values ​​and unrewarded by the environment. The (learning) machine tries to detect in the input data patterns that differ from the structureless noise. An artificial neural network is based on the similarity to the input values ​​and adapts the weights accordingly. It can be learned different things. Popular are the automatic segmentation ( clustering) or the compression of data dimension reduction.

Segmentation

Here similar patterns are represented by a segmentation similar segments.

A very simplified example: you different fruits Imagine (apples, pears, strawberries, oranges), all located in the same basket. So The basket includes the quantity of " segmented " data. Now, a fruit is indiscriminately remove. Then similarities are to be found with the existing on the ground fruits. If something suitable is found, the fruit is said to be defined. If not, then you put it somewhere where space is. Thus, as long as continue until all the fruit is "segmented " according to their characteristics (appearance, smell, color, taste, etc.) were. On the floor are now several piles of fruit, sometimes greater than, less than or equal depending on the frequency of occurrence. These are all practical purposes the cluster.

Compression

An attempt is made to represent many input values ​​in a more compact form, with as little information is lost. Principal component analysis may be, for example, understood as the compression method, when the least significant components of the data is omitted. This corresponds to a practically linear AutoEncoder; This is a multi-layered artificial neural network, the target values ​​are the input values, where a hidden layer of nodes is less than input values ​​as a "bottleneck". The activation of these neurons are the compressed data, from which as much as possible to reconstruct the original data ( uncompressed ) should be.