Problem format:
Row 1 contains n and p.
Row 2i contains a_{i} and the number of "nonzeros" in z_{i};
row 2i+1 contains the indices and the values of the nonzeros in z_{i},
i=1,...,n. (Some of the nonzeros actually have value zero. This is an artifact
of the data set. You are free to remove the zero entries. In matlab, you can use
the command importdata to read in the data files into a matrix with 2 columns.)
These data sets are (after reformatting) from
the libsvm data set .
If you are interested to learn more about current research on methods for SVM, see
the introduction of a paper I recently wrote with Sangwoon Yun
cgd_svm.pdf