Applying Fuzzy Logic to Neural Modeling.
Sébastien Hélie, Sylvain Chartier, Robert Proulx · 2004
What is fuzzy logic? Fuzzy logic is an extension of Boolean logic which allows intermediate values between True and False. As in Boolean logic, a true statement is expressed by the value “1 ” and a false statement by the value “0”. However, unlike in probability theory, the value must not be interpreted as a confidence level but rather as a Membership Function (MF). Therefore, every statement is “True ” to a certain degree and “False ” to another. An interesting property of these MFs is that, because they vary between zero and one, they can be manipulated like probabilities, even though they are interpreted differently. Fuzzy logic, categorization and neural modeling Early results in psychology suggest that humans are using fuzzy decision rules. For example, Rosch (1973) has argued in favor of fuzzy boundaries between natural categories. A strong evidence in favor of fuzzy decision rules is seen in an experiment by Labov (1973) which showed that the