Fig. 6: Scree plot and classification performance using different dimension reduction techniques.

a Cumulative variance explained by each dimension of PCA, NLPCA, and Isomap, averaged across the down-sampled data of eight subjects. b Mean grasp classification performance after progressively removing dimensions from high variance to low, for each of the three dimensionality reduction techniques. Results are averaged across eight subjects with five repetitions each. Error bars indicate ± SEM.