Data augmentation for disruption prediction via robust surrogate models

The goal of this work is to generate large statistically representative data sets train machine learning models for disruption prediction provided by from few existing discharges. Such a comprehensive training database important achieve satisfying and reliable results in artificial neural network classifiers. Here, we aim robust augmentation the multivariate time series using Student $t$ process regression. We apply regression state space formulation via Bayesian filtering tackle challenges imposed outliers noise set reduce computational complexity. Thus, method can also be used if resolution high. use an uncorrelated model each dimension impose correlations afterwards colouring transformations. demonstrate efficacy our approach on plasma diagnostics three different classes DIII-D tokamak. To evaluate distribution generated similar data, additionally perform statistical analyses methods analysis, descriptive statistics classic clustering algorithms.