Fig. 1.

An overview of the EBEN algorithm. 1) Initialize model parameters and the statistical model. The unknown parameters μ denotes the mean of phenotype, $\tilde{y}$ denotes the initial dependent variable and ${\sigma }_0^2$ denotes the variance of the model, obtain the initial features satisfying $k={\mathit{arg}}_i\left\{\left|x_i^T\tilde{y}\right|,\forall i\right\}$. Here, k denotes the subscripts of features, x _i denotes the vector of feature i, $\tilde{y}$ denotes the dependent variable in the statistical model, and α _k is a variable calculated from ${\sigma }_k^2$, 2) Update the parameters in the model during iterations, 3) Use t-test to perform hypothesis test on the estimated value, and 4) Output β ^′ that denotes the significant results and the covariance matrix