Automated supervised classification of variable stars I. Methodology

摘要: The fast classification of new variable stars is an important step in making them available for further research. Selection science targets from large databases much more efficient if they have been classified first. Defining the classes terms physical parameters also to get unbiased statistical view on variability mechanisms and borders instability strips. Our goal twofold: provide overview stellar that are presently known, some relevant parameters; use class descriptions obtained as basis automated `supervised classification' databases. Such will compare assign objects a set pre-defined training classes. For every class, literature search was performed find many well-known member possible, or considerable subset too were present. Next, we searched on-line private their light curves visible band period analysis harmonic fitting. derived curve used describe define classifiers. We compared performance different classifiers percentage correct identification, confusion among computation time. how well can be separated using proposed future improvements made, based such assembled by CoRoT Kepler space missions.