Machine learning techniques for gravitational waves data analysis

In this work we, analysed the capability of machine learning in boosting the MBTA’s pipeline detection capabilities using a random forest algorithm. The results seem to be encouraging in exploring this new tool for the CBC detection’s pipeline MBTA. The machine obtains a detection statistic that is compatible or superior with respect to MBTA. This results in increasing the detectable distances, as stated by fig. 3. In any cases, further studies to increase the interpretability of the detection statistics must be performed. In order to build a tool that is effectively useful in the detection, we must besure how the features and the hyper-parameters impact the model, also avoiding overfitting is fundamental to have a reliable machine. Also, a grid or random search in order to find the best hyper-parameters combination will be helpful for optimizing the machine. A possible perspective for the future may be to define a new ranking statistic based on this machine learning approach.