Title:Modeling quasar variability through self-organizing map-based process

Abstract:Conditional Neural Process (QNPy) has shown to be a good tool for modeling quasar light curves. However, given the complex nature of the source and hence the data represented by light curves, processing could be time-consuming. In some cases, accuracy is not good enough for further analysis. In an attempt to upgrade QNPy, we examine the effect of the prepossessing quasar light curves via the Self-Organizing Map (SOM) algorithm on modeling a large number of quasar light curves. After applying SOM on SWIFT/BAT data and modeling curves from several clusters, results show the Conditional Neural Process performs better after SOM classification. We conclude that SOM classification of quasar light curves could be a beneficial prepossessing method for QNPy.