Changeset 188


Ignore:
Timestamp:
Oct 20, 2004, 5:54:56 PM (18 years ago)
Author:
Peter
Message:

docs

File:
1 edited

Legend:

Unmodified
Added
Removed
  • trunk/src/SVM.h

    r167 r188  
    7272    ///
    7373    /// Training the SVM following Platt's SMO, with Keerti's
    74     /// modifacation. However the complete kernel is stored in memory. The
    75     /// reason for this is speed. When number of samples N is large this is
    76     /// not possible since the memory cost for the kernel scales N^2. In that
    77     /// case one should follow the SMO and calculate the kernel elements
    78     /// sequentially
     74    /// modifacation. However the complete kernel is stored in
     75    /// memory. The reason for this is speed. When number of samples N
     76    /// is large this is not possible since the memory cost for the
     77    /// kernel scales N^2. In that case one should follow the SMO and
     78    /// calculate the kernel elements sequentially. Minimizing \f$
     79    /// \frac{1}{2}\sum
     80    /// y_iy_j\alpha_i\alpha_j(K_{ij}+\frac{1}{C_i}\delta_{ij}) \f$,
     81    /// which corresponds to minimizing \f$ \sum w_i^2+\sum
     82    /// C_i\xi_i^2 \f$
    7983    ///
    8084
Note: See TracChangeset for help on using the changeset viewer.