Algorithm 2: LogitBoost

Input: Dataset (Training)

1. Initialize: $b^{\left(i\right)\ast }$ = (b(i) + 1)/2, $w_i=\frac{1}{n},$ G($a^{\left(i\right)}$) = 0, and p($a^{\left(i\right)}$) = 0.5

2. For time = 1…..,T

2.1a Calculate the working response as well as weights

$z_i$ = $\frac{b^{\left(i\right)\ast }-\mathrm{p}\left(a^{\left(i\right)}\right)}{\mathrm{p}\left(a^{\left(i\right)}\right)\left(1-\mathrm{p}\left(a^{\left(i\right)}\right)\right)}$,

$w_i=\mathrm{p}\left(a^{\left(i\right)}\right)\ast \left(1-\mathrm{p}\left(a^{\left(i\right)}\right)\right)$

2.1b Fit $g_1\left(a\right)$ using weights $w_i$

2.1c Update G($a^{\left(i\right)}$) ← G($a^{\left(i\right)}$) + $\frac{1}{2}{g}_t\left(a^{\left(i\right)}\right)$ and $\mathrm{p}\left(a^{\left(i\right)}\right)=\frac{\exp \left(\mathrm{G}\left(a^{\left(i\right)}\right)\right)}{\exp \left(\mathrm{G}\left(a^{\left(i\right)}\right)\right)+ \exp \left(-\mathrm{G}\left(a^{\left(i\right)}\right)\right)}$

Output: G(a) = ${{\displaystyle \sum }}_{t=1}^T{g}_1\left(a\right)$