The Occam's razor principle for degenerate selfguessing fuzzy classifiers

Let G/sub 1/, C/sub 2//spl epsiv/DSg (/spl Theta/) be arbitrary degenerate self-guessing (for fixed learning set /spl Theta/) fuzzy classifiers. We'll say that G/sub 2/ is more effective than G/sub 1/ iff the performance of G/sub 2/ is higher than G/sub 1/. The problem of choice the "best" fuzzy classifier from DSg (/spl Theta/) set is considered. The unambiguous concept of simplicity and the possible interpretation of the simplicity (Occam's razor) principle for degenerate fuzzy classifiers are introduced. The quality criterion for fuzzy classifiers from DSg (/spl Theta/) is proved.