Algorithm 1.

AdaBoost

1. Initialize weights distribution of training samples:    D1 = (ω11, ω12, ...ω1i..., ω1N), ω1i = 1/N, i = 1, 2...., N

2. For m = 1, 2, ⋯ , M multiple iterations:    (1) Training the weighted Dm sample set to obtain the base learner Gm(x)    (2) Calculating the maximum error of the training set: Em = max∣yi − Gm(xi)∣    (3) Calculating the relative error of each sample:    $e_{mi}=\frac{\mid y_i-G_m(x_i)\mid }{E_m}$    (4) Calculating the regression error rate:    $e_m=\sum _{i=1}^N{\omega }_{mi}{e}_{mi}$    (5) Calculating the weight coefficients of weak learners:    ${\alpha }_m=\frac{e_m}{1-e_m}$    (6) Updating the weight distribution of the sample set:    ${\omega }_{m+1,i}=\frac{{\omega }_{mi}}{Z_m}{{\alpha }_m}^{1-e_m}$    $Z_m=\sum _{i=1}^N{\omega }_{mi}{\alpha }_m^{1-e_{mi}}$    Dm+1, i = (ωm+1, 1, ωm+1, 2, ...ωm+1, i..., ωm+1, N)

3. Output the ultimate strong learner    $f\left(x\right)={\displaystyle {\sum }_{m=1}^M(\ln \frac{1}{{\alpha }_m}){\alpha }_m{G}_m\left(x\right)}$ end