Online Boosting Algorithm Based on Two-Phase SVM Training

Figure 2

Illustration of the convergence process: (a) the first linear classifier trained (𝑖=50, 𝑇=1), (b) several additional weak classifiers added (𝑖=250, 𝑇=5), (c) the data separation corresponding to the set of classifiers in (b). Background color shows class generated by classifier, form and color of data points show actual label 𝑦, (d) data separation achieved after convergence (𝑖=5000, 𝑇=100).

(a)

(b)

(c)

(d)