www-ai.cs.tu-dortmund.de/LEHRE/FACHPROJEKT/SS14/Papers/libsvm.pdf
warning message to indicate that the code may be faster without
shrinking.
5.7 Computational Complexity
While Section 4.1.7 has discussed the asymptotic convergence and the local conver-
gence rate of [...] caching. LIBSVM provides some special settings for unbalanced data; details are
in Section 6. Section 7 discusses our implementation for multi-class classification.
Section 8 presents how to transform SVM [...] i = 1, . . . , l.
The dual problem is
min α
1
2 αTQα
subject to 0 ≤ αi ≤ 1/(νl), i = 1, . . . , l, (7)
eTα = 1,
where Qij = K(xi,xj) = φ(xi) Tφ(xj). The decision function is
sgn
( l∑
i=1
αiK(xi,x)− ρ
) …