Testing of Feature Extracting Methods on Bar Problem
Abstrakt
Compared is performance of several dimension reduction techniques as a tool for feature extraction. Studied are namely singular value decomposition, FastMap, semi-discrete decomposition, non-negative matrix factorization, novel neural network based algorithm for Boolean factor analysis and three frequently used cluster analysis methods as well. So called bars problem is used as the benchmark. Set of artificial signals generated as a Boolean sum of given number of bars is analyzed by these methods. Resulting images show that Boolean factor analysis is upmost suitable method for this kind of data.