Charged-particle identification with advanced artificial intelligence approaches

Modern nucleus-nucleus collision experiments require the use of advanced particle identification techniques. However, similar tasks are often timeconsuming, enhancing the complexity of the data analysis process. We develop a novel approach capable to automatically identify charge and mass of detected ions with almost zero human supervision. Our method uses evolutionary computing and clustering algorithms and exploits previously developed analytical functionals to provide physics constraints. The new algorithm is successfully tested on Delta E-E telescopes based on annular silicon strip detectors and could be integrated in online and ovine analysis pipelines of existing detection arrays.