Fig. 1.

Tuning and comparison to address the data nonlinearity. (A) TE for LLE (yellow line), LTSA (blue line) and Isomap (green line). The x-axis reports different values of neighborhood parameter k; y-axis reports the index for TE. (B) Classification accuracy for LLE (yellow line), LTSA (blue line) and Isomap (green line). The x-axis reports different values of neighborhood parameter k; y-axis reports values of accuracy. (C) Best performance in linear discrimination compared between tested algorithms.