Skip to main content

Table 3 Prediction performance in terms of root mean square error of prediction (RMSE) of the four models (MLP_20, LR, MLP_1, RT) reported by Khaki and Wang (2019); A) in a maize dataset. MPL_20 denotes the MLP model with 20 hidden layers, and MPL_1 denotes the MLP model with 1 hidden layer. Prediction performance in terms of Pearson’s correlation of 6 species across traits evaluated with 11 methods (Azodi et al., 2019); B). SVR denotes support vector regression. SVR_lin denotes SVR with linear kernel, SVR_poly denotes SVR with polynomial kernel, SVR_rbf denotes SVR with kernel Radial Basis Function

From: A review of deep learning applications for genomic selection

A

 

Model

Trait

RMSE

  
  

MLP_20

Yield

12.79

  
   

Check yield

11.38

  
   

Yield difference

12.4

  
  

LR

Yield

21.4

  
   

Check yield

19.87

  
   

Yield difference

13.11

  
  

MLP_1

Yield

18.04

  
   

Check yield

15.18

  
   

Yield difference

15.19

  
  

RT

Yield

15.03

  
   

Check yield

14.87

  
   

Yield difference

15.92

  

B

Method

Maize

Rice

Sorghum

Soy

Spruce

Switch-grass

 

rrBLUP

0.44

0.34

0.63

0.46

0.32

0.61

 

BRR

0.44

0.39

0.63

0.46

0.32

0.61

 

BayesA

0.42

0.38

0.63

0.47

0.32

0.61

 

BayesB

0.43

0.38

0.63

0.46

0.32

0.61

 

BL

0.44

0.39

0.62

0.46

0.32

0.61

 

SVR_lin

0.41

0.38

0.62

0.43

0.19

0.6

 

SVR_poly

0.43

0.38

0.63

0.41

0.33

0.61

 

SVR_rbf

0.39

0.38

0.63

0.04

0.34

0.6

 

RF

0.43

0.4

0.58

0.36

0.35

0.57

 

GTB

0.37

0.38

0.58

0.4

0.33

0.56

 

MLP

0.17

0.08

0.45

0.44

0.28

0.45