Experimental study in emotion recognition using prosodie features

Respiratory Sound Classification for COVID-19 Preliminary Screening

“…The metrics being used in this paper will be the following: accuracy, precision, recall, and F1-score [22]. These metrics are defined by the equations below: where TP represents all the correctly labeled positive diagnosis classification labels, TN represents all of the correctly labeled negative diagnosis classification labels, FP represents all the wrongly labeled positive diagnosis classifications labels and FN represents all the wrongly labeled negative diagnosis classifications labels.…”

“…Macro-average will compute the metric independently for each class and take the total average, thereby treating all classes equally, whereas a micro-average will aggregate the contributions of all classes to compute the average metric. Typically, in an unbalanced dataset, it is more advantageous to consider the micro-average above the macro-average as it is more favourable to classes that have fewer samples [22]. Finally, weighted average calculates each class independently and aggregates them using a weight which depends on the quantity of true labels for each class present.…”