Table 2.

Results of the fourfold cross validation using the DNN, CNN and CNN-LSTM architectures for 4 mixture gases

4-fold cross validation for 4 mixture gases		Set 1	Set 2	Set 3	Set 4	Average
DNN	Accuracy	99.3%	99.3%	99.3%	99.3%	99.3%
	Loss	0.023	0.023	0.094	0.022	0.041
	Training Time	25.5 s	25.8 s	1 min 7 s	22.8 s	35.3 s
CNN	Accuracy	99.3%	100%	99.3%	100%	99.7%
	Loss	0.041	0.064	0.054	0.028	0.047
	Training Time	0.9 s	0.8 s	1.3 s	0.9 s	1.0 s
CNN-LSTM	Accuracy	99.3%	100%	100%	100%	99.9%
	Loss	0.016	0.0016	0.0016	0.0026	0.013
	Training Time	1.6 s	1.6 s	1.6 s	1.6 s	1.6 s