Skip to Main Content
Machine learning tools have been used since a long time ago to study disruptions and to predict their occurrence. On the other hand, the challenges posed by the quality and quantities of the data available remain substantial. In this paper, methods to optimize the training data set and the potential of kernels-based advanced machine learning tools are explored and assessed. Various alternatives, ranging from appropriate selection of the weights to the inclusion of artificial points, are investigated to improve the quality of the training data set. Support vector machines (SVM), relevance vector machines (RVMs), and one-class SVM are compared. The relative performances of the different approaches are initially assessed using synthetic data. Then they are applied to a relatively large database of JET disruptions. It is shown that in terms of final results, the optimization of the training databases proved to be very productive. Further, the RVM algorithm performs well when it is trained on a small set of discharges compared to the traditional methods.