Gait Biomarkers Classification by Combining Assembled Algorithms and Deep Learning: Results of a Local Study

Table 5

Results of combining machine learning algorithms.

Metaclassifier	Deep learning	10-fold cross-validation	2/3–1/3	Representative sample
	Dl4jMlpClassifier	75.7	79.05	80.1105
AdaBoostM1	Dl4jMlpClassifier	76.6333	75.5	80.3867
AttributeSelectedClassifier	Dl4jMlpClassifier	77.3833	77.4	78.1768
Bagging	Dl4jMlpClassifier	79.4167	79.65	81.4917
ClassificationViaRegression	Dl4jMlpClassifier	66.6667	67.5	69.8895
CVParameterelection	Dl4jMlpClassifier	75.7	79.05	80.1105
FilteredClassifier	Dl4jMlpClassifier	81.7333	84.5	85.0829
LogitBoost	Dl4jMlpClassifier	66.6667	67.5	69.8895
MultiClassClassifier	Dl4jMlpClassifier	75.7	79.05	80.1105
MultiSearch	Dl4jMlpClassifier	75.7	79.05	80.1105
OrdinalClassClassifier	Dl4jMlpClassifier	75.7	79.05	80.1105
RandomCommittee	Dl4jMlpClassifier	79.1833	81.3	80.1105
RandomizableFilteredClassifier	Dl4jMlpClassifier	73.9	76.75	79.8343
RandomSubSpace	Dl4jMlpClassifier	79.7333	80.7	78.7293
Stacking	Dl4jMlpClassifier	36.6667	67.5	69.8895
ThresholdSelector	Dl4jMlpClassifier	72.9167	76.65	47.7901
WeightedInstancesHandlerWrapper	Dl4jMlpClassifier	75.7	79.05	80.1105