Skip to main content
Fig. 3 | BioData Mining

Fig. 3

From: Gene function finding through cross-organism ensemble learning

Fig. 3

Evaluation results by varying the likelihood threshold. Results obtained by varying the value of the likelihood threshold and using Random Forest as supervised algorithm, trained on perturbed versions of the Homosapiens annotation matrix with perturbation percentage p=10%. SM is the single model method, proposed in [57]; AVG considers the average of the likelihood scores given by the models inferred from five different perturbation random seeds; ∩1/5 considers the union of the predictions from the five models; ∩5/5 considers the intersection of the predictions from the five models; ∩3/5 considers those predictions from three out of the five models

Back to article page
\