Skip to main content

Table 1 The summary of the parameterization of used architectures — C–D–GEX, CR–D–GEX, T–D–GEX, and TR–D–GEX do not differ in number of parameters and neurons. Note that the total number of parameter remains approximately the same across architectures

From: On tower and checkerboard neural network architectures for gene expression inference

Towers Neurons/tower Neurons Parameters
1 10,000 34,759 257,149,036
2 7,227 48,121 257,119,561
3 5,941 58,228 257,068,283
4 5,157 66,643 256,997,503
5 4,615 73,984 256,977,621
6 4,211 80,557 256,953,201
8 3,637 92,047 256,729,407
10 3,242 102,019 256,587,674
12 2,948 110,887 256,374,168