Skip to main content
. 2018 May 18;18(5):1615. doi: 10.3390/s18051615

Table 2.

Classification performance of 8 multi-feature sets using an SVM classifier when the training and testing sets are acquired from multiple conditions for multiple EMG datasets (Datasets 1–3).

Multi-Feature Set Able-Bodied Subjects Amputee Subjects Mean
Limb Position Forearm Orientation Contraction Intensity Contraction Intensity
1000 200 d 1000 200 d 1000 200 d 1000 200 d 1000 200 d
MS1 (TD) 98.3 93.9 1.7 * 98.2 92.7 1.3 * 99.1 96.8 1.0 * 89.8 78.4 1.6 * 96.3 90.4 1.4
MS2 98.8 93.3 2.0 * 98.7 92.0 1.5 * 99.3 96.2 1.2 * 91.9 78.7 2.1 * 97.2 90.1 1.7
MS3 99.1 94.1 2.1 * 99.1 92.4 1.6 * 99.5 96.7 1.2 * 93.5 79.9 2.2 * 97.8 90.8 1.8
MS4 97.1 84.6 3.5 * 96.9 84.8 2.2 * 97.7 88.6 1.8 * 88.5 72.1 2.2 * 95.0 82.5 2.4
MS5 98.9 93.4 2.1 * 98.7 91.6 1.6 * 99.4 95.6 1.3 * 92.7 77.6 2.4 * 97.4 89.6 1.8
MS6 98.6 92.6 2.1 * 98.1 89.6 1.9 * 99.0 94.0 1.5 * 91.9 76.1 2.6 * 96.9 88.1 2.0
MS7 97.1 93.8 1.2 * 97.6 92.8 1.2 * 98.6 96.2 0.9 * 85.9 78.8 0.9 * 94.8 90.4 1.0
MS8 99.1 93.7 2.1 * 99.0 92.2 1.6 * 99.6 96.4 1.4 * 93.1 79.6 2.1 * 97.7 90.5 1.8

* Denotes a significant difference between 1000 Hz and 200 Hz (p<0.05).