Skip to main content
. 2010 Apr 20;4(4):295–313. doi: 10.1007/s11571-010-9110-4

Fig. 1.

Fig. 1

Data for training and testing the PU (processing unit) in Example 3 and Example 4 are shown as the vertices of a cube. Their bipolar binary labels are the numbers or question marks for unknown labels in the squares at the vertices. x4, x5, x6 are unavailable in the data set for Example 3. They are learned one by one without supervision (i.e., with labels generated by the PU) in Example 4