Distinguishing alpha and beta signals in silicon detectors is important input to accurate event reconstruction in experiments such as the Beta-decay Paul Trap. In this work we used basic machine-learning techniques to develop a simple, high-accuracy discrimination model.