Table 4 Six-sensor model classification results. Bold lines indicate best-performing networks.
From: Detection of balance disorders using rotations around vertical axis and an artificial neural network
Number of hidden layer neurons | AlgorithmA | Number of epochs | Accuracy (%) | Sensitivity (%) | Specificity (%) |
|---|---|---|---|---|---|
4 | RMSprop 0.01, 0.7 | 100 | 78 | 83 | 73 |
4 | RMSprop 0.01, 0.7 | 200 | 84 | 83 | 84 |
4 | RMSprop 0.01, 0.7 | 400 | 82 | 83 | 82 |
5 | RMSprop 0.01, 0.7 | 50 | 81 | 81 | 82 |
5 | RMSprop 0.01, 0.7 | 100 | 86 | 88 | 84 |
5 | RMSprop 0.01, 0.7 | 200 | 74 | 69 | 78 |
5 | RMSprop 0.01, 0.7 | 1000 | 79 | 83 | 76 |
5 | RMSprop 0.001, 0.7 | 100 | 80 | 81 | 80 |
5 | RMSprop 0.001, 0.7 | 200 | 84 | 81 | 86 |
5 | RMSprop 0.001, 0.7 | 400 | 80 | 81 | 80 |
6 | RMSprop 0.01, 0.7 | 50 | 80 | 81 | 80 |
6 | RMSprop 0.01, 0.7 | 100 | 76 | 76 | 76 |
6 | SGD 0.01, nesterov | 200 | 84 | 83 | 84 |
6 | SGD 0.01, nesterov | 400 | 80 | 81 | 80 |
