. Author manuscript; available in PMC: 2016 Oct 1.

Published in final edited form as: Comput Speech Lang. 2015 Jan;29(1):132–144. doi: 10.1016/j.csl.2014.02.001

Table 6.

Unweighted (weighted) average recall in % of final systems (by chance: 50.0% for unweighted, 64.4% for weighted) on the test set of pathological speech challenge database (NKI-CCRT data).

System	Accuracy (%)

Baseline SVM	68.0(66.2)
Baseline Random Forest	68.9(67.5)

Subsystem fusion (KNN’s posteriors, no smoothing)	70.0(69.2)
Subsystem fusion (SVM’s posteriors, no smoothing)	66.9(67.8)
Subsystem fusion (smoothed posteriors of KNN)	70.6(73.8)
Subsystem fusion (smoothed posteriors of SVM)	73.5(72.8)