<BODY bgcolor="#aaeebb"> <CENTER> <H3>Relationship to Fuzzy Sets</H3> </CENTER> <UL> <LI><B>Softening the objective</B> <UL> <LI>trade the &quot;best for sure&quot; with the &quot;:good Enough with high probability&quot; <BR><BR> </UL> <LI><B>Softening the definitions</B> <UL> <LI>blurring the boundary of &quot;Good Enough&quot; or &quot;High Probability&quot;<BR><BR> </UL> <LI><B>Merging the Qualitative to Quantitative</B> </UL> <BR> <tt> <A HREF="Slide16.html" target="_parent">To Next Slide</A> / <A HREF="Slide14.html" target="_parent">To Previous Slide</A> / <A HREF="../OOTOC.html" target="_parent">Table of Content</A> </tt> <BR> <hr> <hr> <br> <P> It is also appropriate at this point to address the relationship between OO and Genetic Algorithms (GA) and to Fuzzy Logic (FL). In a sentence, OO is completely complementary to GA and FL. While OO deals with softening the goal of optimization, FL can further blur the definitions used for softening. For example, we may say that the top-5% represents the &quot;good enough&quot;. But what about the top-6%? Are they necessarily bad? A more reasonable definition of good enough can be defined using the membership function of fuzzy logic permitting a gradual degradation. Similarly, high probability need not be defined by a sharp boundary of &gt;0.95. Instead, a linearly declining membership from 1 to 0 for P=0.95 to 0.9 can be used. The point is that all the material we have discussed w.r.t. OO can be and probably should be fuzzified. There is no conflict with FL. <BR> <BR> </BODY> </HTML>