Features extraction applied to the difficult data sets description on the bioinformatics example

In this article, we consider the problem of extraction of vector features as essential and important for the object descriptions, especially in original Big Data. This problem concerns both the real properties of signals and images as well as simulated and artificial ones. Search methods are often difficult to identify due to data that is characterized by excessive complexity and redundancy.

To find hidden attributes we use a special algorithm to estimate the density of feature vectors. Finally, this procedure generates a sufficiently small set of features that properly describe the features of objects. The content of the resultant subset may vary, depending on the defined goal.

Author: Artur WiliƄski
Conference: Title