Abstract :
The complexity of the switching networks necessary to realize arbitrary combinational functions is studied. Asymptotic upper and lower bounds for fully defined functions are well known, while lower bounds also exist for partially defined combinational functions. The present paper supplements these results with the upper bounds for the partially defined functions. The results have possible relevance to pattern recognition.
Keywords :
Complexity bounds, limited fan-in, partially defined switching functions, threshold gate networks.; Neural networks; Pattern recognition; Upper bound; Complexity bounds, limited fan-in, partially defined switching functions, threshold gate networks.;
