Fig. 10From: Co-MLHAN: contrastive learning for multilayer heterogeneous attributed networksTesting and validation F1 macro for the final classifier with early-stopping (left) and without early-stopping (right)Back to article page