... it¹

These datasets are: arrythmia, audiology, bach chorales, echocardiogram, isolet, mobile robots, waveform.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... ten-fold²

For consistency with previous results, 5-fold cross validation was used for the Auslan domain.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... level³

All significance statements are made using a paired t-test.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... binarizing⁴

That is, the classification problem is turned into a binary one of classifying instances as either X or not-X.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... off-line⁵

Off-line Chinese character recognition is where one is presented with images of handwritten characters. On-line Chinese handwriting recognition is where one observes the character being drawn and hence observes the order of the strokes. Off-line is considered more difficult.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.