how to use kmeans to blindly split epochs into 2 groups using all features from a given file.

Enclosed is also ‘p3simple.mat’ containing 10 training sets of 20- trial averaged P300 data in trainEpochs. Each row contains one averaged P300 trial and each column corresponds to a 100ms average section of the data. The trainTargets variable identifies whether the recorded ERP was a target while trainCodes identifies which icon was flashing. The remaining variables epochs and codes correspond to ERPs to which the target is unknown, containing 100ms averages and icon numbers respectively. Out of these, please let me know how to use kmeans to split the epochs into two labeled groups (say label 1 and label 2).