Table 1 The prediction accuracy of the subject-classification task, reported as pairs of mean (standard deviation) in percent.

From: Explaining the unique nature of individual gait patterns with deep learning

Model Ground
Reaction
Forces [%]
Joint Angles
Full-Body [%]
Joint Angles
Full-Body
(flex.-ext.) [%]
Joint Angles
Lower-Body [%]
Joint Angles
Lower-Body
(flex.-ext.) [%]
Linear (SGD) 95.4 (1.7) 100.0 (0.0) 96.3 (1.8) 100.0 (0.0) 91.5 (2.2)
Linear (SVM) 100.0 (0.0) 100.0 (0.0) 99.7 (0.4) 100.0 (0.0) 99.8 (0.6)
MLP (3, 64) 88.3 (3.7) 99.9 (0.3) 84.3 (3.3) 99.9 (0.3) 68.5 (8.3)
MLP (3, 256) 96.6 (0.8) 100.0 (0.0) 95.6 (2.6) 100.0 (0.0) 89.4 (4.3)
MLP (3, 1024) 98.8 (1.3) 100.0 (0.0) 97.8 (1.0) 100.0 (0.0) 96.5 (1.2)
CNN-A 99.1 (0.8) 100.0 (0.0) 95.6 (1.7) 99.9 (0.3) 92.0 (3.9)
CNN-C3 99.2 (0.6) 100.0 (0.0) 97.7 (1.5) 100.0 (0.0) 97.0 (1.3)