Skip to main content
Fig. 4 | BMC Bioinformatics

Fig. 4

From: Deep GONet: self-explainable deep neural network based on Gene Ontology for phenotype prediction from gene expression data

Fig. 4

Illustration of the procedure to evaluate the biological significance of neurons from the first layer. The upper part shows how to calculate the rank of the target mask of each neuron. The lower part shows how to compute the rank of the neurons according to their LRP relevance. The relationship between these two metrics is evaluated through a final figure (e.g., Fig. 5)

Back to article page