Table 10.

Performance metrics achieved using augmenting the best-performing weakly labeled augmented training data with one of the COVID-19 CXR collections toward classifying another COVID-19 CXR collection as belonging to the viral pneumonia category.

Dataset	Accuracy
	Twitter-COVID-19	Montreal-COVID-19
Baseline	0.2885	0.5028
Baseline + CheXpert	0.5555	0.6536
Baseline + CheXpert + Twitter	-	0.7095
Baseline + CheXpert + Montreal	0.8889	-

Bold numerical values denote superior performance.