In recent years different methods of ML were considered to detect abnormal TL glow curves for TL dosimeter. Abnormal GC can lead to wrong dose estimation if not properly detected. ML methods could make identification of abnormal GC more robust and at lower cost than using trained technician.

Among the forest of ML methods what would be the most performant methods that could be used for that purpose, knowing that quite large of database of classified GC can be used (about severals thousands of GC). Random forest (RF), artificial neural network (ANN) or support vector machine (SVM) methods?