Figure 6.

Parameter sensitivity analysis. STGRNS is robust to various parameters including dropout, learning rate, epoch, window size, head, and batch size. Performance is measured as AUROC and AUPRC. (a) The parameter for dropout ranges from 0.00 to 0.99 and each increment is 0.01. (b) The parameter for learning rate ranges from 0.0001 to 0.0059 and each increment is 0.0001. (c) The parameter for epoch ranges from 10 to 700 and each increment is 10. (d) The parameter for window size ranges from 10 to 790 (the number of cells in this dataset is 790) and each increment is 10. (e) The parameter for the head ranges from 1 to 200. (f) The parameter for batch size ranges from 1 to 2041 and each increment is 15