Table 1. The confusion matrix of classification results using the 5-fold cross-validation of the CNN algorithm.

		Target class			Sum by row
		Insertion	Withdrawal	Stop
Output class	Insertion	116,437	1,533	1,379	119,349
	Withdrawal	1,586	107,336	1,757	110,679
	Stop	3,484	4,603	90,812	98,899
Sum by column		121,507	113,472	93,948	328,927