Assessing the Effect of Crossing Databases on Global and Local Approaches for Face Gender Classification

Iman-Davenport's Statistic (FF ) is higher than the corresponding value of the F-distribution when statistical di erences are found. Holm's Method: The classi cation models above the double line performed signi cantly worse than the most signi cant model (marked in bold at the bottom) with a 95% signi cance level. Wilcoxon's Test: The symbol • indicates that the classi cation model in the row signi cantly outperforms the model in the column, and viceversa for the symbol ◦ (above the main diagonal with a 90% con dence level, and below it with a 95%). Methodology