Machine Learning In Astrophysical Data

Astronomical surveys provide an ever-increasing amount of data that requires time consuming analysis. There are many parameters that can be used to distinguish different types of astronomical objects. Therefore, statistical tools are increasingly used for this purpose. For this reason, it is very important to automate this process. One possibility is to use supervised learning algorithms. We present an exemplary application of such an algorithm: supervised learning algorithm based on support vector machines (SVM) applied for classification of the WISE data. Machine learning algorithms can also have other uses, for example, to study clustering of the data. For instance, to study galaxy properties and evolution it would be advisable to categorize them into groups with similar properties. This is usually done in low dimensional parameter space, i.e. based on color-color plots making use of only three or four properties. The disadvantage of such a solution is the possibility of overlooking subtle differences between groups of galaxies. For this reason, we have attempted to use unsupervised learning algorithms to divide data in the multidimensional parameter space; we present preliminary results of such a classification performed on the data from the VIPERS survey.