Table 10 Recognition rates for exercise classification on the K3Da dataset using CNN architectures.
From: Spike train analysis in rehabilitation movement classification using deep learning approach
CNN architecture | Accuracy (mean ± std) | 95% CI | Macro-F1 (mean ± std) | Training time (min/epoch) |
|---|---|---|---|---|
CNN1 | 0.9618 ± 0.005 | [0.954–0.969] | 0.958 ± 0.006 | 3.2 |
CNN2 | 0.9432 ± 0.006 | [0.935–0.952] | 0.939 ± 0.008 | 3.6 |
CNN3 | 0.9598 ± 0.004 | [0.952–0.966] | 0.955 ± 0.006 | 2.5 |
CNN4 | 0.9771 ± 0.003 | [0.972–0.982] | 0.973 ± 0.005 | 5.9 |
Proposed CNN | 0.9821 ± 0.002 | [0.978–0.986] | 0.979 ± 0.003 | 4.1 |
