1: \begin{abstract}

2: 

3: 

4:      We present a scalable and effective classification model

5:      to train multi-class boosting for multi-class classification

6:      problems.

7: %

8: %

9: %

10: %

11: %

12: %

13: %

14: %

15:     Shen and Hao introduced a direct formulation of multi-class boosting in the

16:     sense that it directly maximizes the multi-class margin \cite{Shen2011Direct}.

17:     %

18:     %

19:     %

20:     %

21:     The major problem of their approach is its high

22:     computational complexity for training, which hampers its application on

23:     real-world problems.

24:     In this work, we propose a scalable and simple stage-wise

25:     multi-class boosting method, which

26:     also directly maximizes the multi-class margin.

27:     Our approach offers a few advantages:

28:     1)

29:     it is simple and computationally efficient to train.  The approach

30:     can speed up the training time by more than two orders of

31:     magnitude without sacrificing the classification accuracy.

32:     2)

33:     Like traditional AdaBoost, it is less sensitive to the

34:     choice of parameters and empirically demonstrates excellent

35:     generalization performance.

36: %

37: %

38: %

39: %

40:     Experimental results on challenging multi-class machine

41:     learning and vision tasks demonstrate that the proposed approach

42:     substantially improves the convergence rate and accuracy of the

43:     final visual detector at no additional computational cost

44:     compared to existing multi-class boosting.

45: 

46: \end{abstract}

47: