Skip to main content

Table 1 Hyper-parameters of proposed deep architecture

From: An Ensemble Deep Learning based Predictor for Simultaneously Identifying Protein Ubiquitylation and SUMOylation Sites

Subnets Layer category Hyper-parameters
Activation function Size Filters Dropout
Sequence 1D Convolution Relu 2 201 0.4
Relu 3 151 0.4
Relu 5 101 0.4
Dense Relu 256 0.3
Relu 128 0
Sigmoid 2
Physico-O Dense Relu 256 0.2
Relu 128 0.1
Sigmoid 2
Physico-P Dense Relu 512 0.3
Relu 256 0.2
Relu 128 0.1
Sigmoid 2
Physico-H Dense Relu 1024 0.4
Relu 512 0.3
Relu 256 0.2
Relu 128 0.1
Sigmoid 2
Physico-C 1D Convolution Relu 2 201 0.2
Relu 3 151 0.1
Dense Sigmoid 2
Physico-B 1D Convolution Relu 2 201 0.3
Relu 3 151 0.2
Relu 5 101 0.1
Dense Sigmoid 2
Physico-A 1D Convolution Relu 2 201 0.4
Relu 3 151 0.3
Relu 5 101 0.2
Relu 7 51 0.1
Dense Sigmoid 2
Ensemble Dense Relu 7
Sigmoid 2