Table 8.
The results obtained for the whole signals during experiment 3 for various segment sizes and both networks. The chaos class is denoted as Positive and non-chaos as Negative.
| Model | Segment | Accuracy | TP | FN | TN | FP |
|---|---|---|---|---|---|---|
| LSTM | 50 | 0.967 | 107,389 | 5,651 | 195,423 | 4,637 |
| 100 | 0.985 | 55,233 | 1,287 | 99,022 | 1,008 | |
| 200 | 0.984 | 27,708 | 552 | 49,352 | 663 | |
| CNN | 50 | 0.924 | 108,288 | 4,752 | 180,873 | 19,187 |
| 100 | 0.902 | 54,781 | 1,739 | 86,382 | 13,648 | |
| 200 | 0.957 | 27,360 | 900 | 47,584 | 2,431 |