Table 1.
Two-dimensional BKS.
Fig 1.
A hierarchical agent-based framework with the BKS.
Table 2.
Prediction outputs of Agents 1 and 2.
Table 3.
Creation of BKS for the classification scenario in Table 2.
Fig 2.
Configuration of the hierarchical agent-based framework used in the experiments.
Table 4.
List and descriptions of benchmark datasets.
Table 5.
Accuracy rates.
Table 6.
F1 scores.
Fig 3.
Number of BKS wins over majority voting in data sets with and without noise (red, yellow, and green lines indicate the threshold of wins requires for significance level of α = 0.1, 0.05 and 0.01, respectively).
Table 7.
Accuracy rates with and without noise.
Table 8.
Comparison of F1 scores with literature (best in bold).
Table 9.
List of features and description.
Fig 4.
Feature importance using DT, RF, and XGBoost.
Table 10.
Accuracy rates and BKS wins with noise added.
Table 11.
F1 scores with noise added.
Table 12.
Accuracy rates with different ratios of minority to majority samples.