when you load 'MINST_data.mat' into matlab workspace, there are four matrices,

'train_samples'            4000x784
'train_samples_labels'     4000x1
'test_samples'             1000x784
'test_samples'             1000x1

Each row in 'train_samples' or 'test_samples' is a sample of a handwritten
digit.  'train_samples_labels' and 'test_samples_labels' gives
corresponding groundtruth labels, '0' to '9'.  Note the number of training
samples for each class is around 400, but not exactly 400.


You can display the i-th image by following matlab command:

imagesc(reshape(train_samples(i,:),[28,28]));

A baseline DAGSVM achieves 95% accuracy with a 6-degree polynomial kernel
function.

Here is the confusion matrix obtained for the baseline DAGSVM.  The
vertical axis is for the ground truth class label, and the horizonatal axis
is for the estimated class label.

    84     0     0     0     0     1     2     0     1     0
     0   121     0     0     0     0     0     0     0     0
     0     0   110     0     1     0     0     0     1     0
     0     0     0   106     0     2     0     1     2     0
     0     0     0     0   102     1     0     3     1     1
     1     0     0     5     0    86     2     0     0     0
     1     0     0     1     1     1    82     0     0     0
     0     0     1     2     0     0     0    93     2     0
     0     1     1     1     0     1     1     1    78     2
     0     0     1     0     4     0     0     1     1    89