Table 4.

Average percent accurate classification for each classification model excluding a set of variables. Values are presented as mean ± standard deviation. L-M = Levenberg-Marquardt algorithm.

Parameter excluded	Multilayer Perceptron LM	Support Vector Machine	Learning Vector Quantization
None	96.44 ± 1.27	85.39 ± 2.4	91.03 ± 0.98
Pressures	86.29 ± 1.49	87.08 ± 1.12	90.73 ± 5.45
Durations	89.22 ± 0.39	90.17 ± 1.69	84.45 ± 3.75
Velopharynx	95.61 ± 3.62	83.43 ± 0.56	90.13 ± 5.46
Tongue base	89.48 ± 0.83	84.55 ± 1.08	85.71 ± 3.49
UES	83.57 ± 2.19	84.55 ± 2.81	79.45 ± 4.35
Integrals	91.84 ± 0.65	86.52 ± 0.92	85.50 ± 1.75
Line integrals	90.25 ± 1.49	83.43 ± 1.92	84.78 ± 5.12