Fig. 4From: Evaluation of maxout activations in deep learning across several big data domainsRectified linear unitBack to article page