EnsembleSplice: ensemble deep learning model for splice site prediction

BMC Bioinformatics

Table 1 EnsembleSplice neural network hyper-parameter search space

Neural Network	Hyper-parameter	Range	Steps	Selected
CNN	Filters	8–400	8	72, 120, 136, 144, 168, 208, 250, 272,
	Kernel size	1–9	2	3, 4, 5, 7, 9
	Dropout	0.05–0.30	0.05	0.20, 0.35
	Max-Pool size	1–9	2	3
DNN	Units	32–704	32	32, 128, 224, 250, 256, 352, 512, 704,
	Kernel regularizers	0.0025, 0.025, 0.036	-	0.0025, 0.025, 0.036
	Dropout	0.05–0.50	0.50	0.1, 0.15, 0.25

This table shows the convolutional neural network (CNN) and Dense Neural Network (DNN) search space. This includes the search range, steps and the selected hyperparameter

ISSN: 1471-2105