TABLE I:

Coordinate prediction accuracy and segment classification of the two presented models versus the three comparison methods as described in section III-C. For coordinate prediction, we report a mean error (in millimeters) and its 95% confidence interval. For segment classification, we report the percentage of correctly classified segments. All results are reported on a separately held-out test set.

Model/Task	Coordinate Prediction (in mm)	Classification (in %)
QRSi	15.09±0.20	47.35
CNN	14.02 ±0.22	54.79
SAE (LSTM)	14.63±0.22	52.16
SAE (GRU)	14.44±0.20	51.14
f-SAE (LSTM)	13.14±0.23	53.29
f-SAE (GRU)	12.84±0.22	56.29