Comparison of Machine Learning-Based Radioisotope Identifiers for Plastic Scintillation Detector

Background: Identification of radioisotopes for plastic scintillation detectors is challenging because their spectra have poor energy resolutions and lack photo peaks. To overcome this weakness, many researchers conducted radioisotope identification studies using machine learning algorithms; however, the effect data normalization on has not been addressed yet. Furthermore, learning-based identifiers are limited.Materials Methods: In study, were implemented, performances according to methods compared. Eight classes consisting combinations 22Na, 60Co, 137Cs, background, defined. The training set was generated by random sampling technique based probabilistic density functions acquired experiments simulations, test experiments. Support vector (SVM), artificial neural network (ANN), convolutional (CNN) implemented as with six methods, trained set.Results Discussion: evaluated sets without gain shifts confirm robustness against shift effect. Among three identifiers, prediction accuracy followed order SVM >ANN>CNN, while time SVM>ANN>CNN.Conclusion: combined highest SVM. CNN exhibited a minimum variation in each class, even though it had lowest among identifiers. sets, its shortest