Supervised Anomaly Detection in Univariate Time-Series Using 1D Convolutional Siamese Networks