Table 9.

Comparative Results using Benchmark Dataset 1 for RF, RNN, CNN-ext. and CNN. RF refers to Random Forest. RNN refers to Recurrent Neural Network. CNN-ext. refers to extended CNN where we use our original architecture with another convolutional layer and max pooling layer adding after the original max pooling layer. CNN refers to the Convolutional Neural Network described in the paper

#	Class/Group	Training				Independent Test
		RF	RNN	CNN-ext	CNN	RF	RNN	CNN-ext	CNN
1	Level 1	0.97	0.43	0.95	0.96	0.95	0.70	0.69	0.70
2	Class A	0.97	0.16	0.97	0.98	0.75	0.70	0.78	0.78
3	Class B/Group 3	0.94	−0.04	0.96	0.96	0.94	0.34	1.00	0.99
4	Class C/Group 1	0.92	0.20	0.96	0.97	0.90	0.54	0.89	0.89
5	Class D	1.00	0.66	0.99	1.00	0.44	0.06	1.00	0.91
6	Group 2	0.96	0.42	0.97	0.97	0.75	0.34	0.81	0.81