Good Solutions will Emerge without a Global Objective Function: Applying Organizational-Learning Oriented Classifier System to Printed Circuit Board Design - 0 Keiki Takadaiiia
Takao Terano · 1997
This paper descrilw a novel rvolut.ionary comput.at,ion iiiotl(4: Or~~nizat.ional-I,eavriit~g Ovien(.c.t-l ClassiJicr Syst.cni (OCS), and it,s applicat~ion t,o l'rint,ed Circuit l3oards ( PCBs) design 1)rol,le1ns. Tlie idra of OC'S conies from t.he theory of Organizational Learning in organiza,t.ioiial sciences. OC'S is aii est.eiided niult.iagciit. version of a conveihond Learning Classifier Syst.em 1.0 learn adapt.ive rulcs ill a given environment.. OCS adaptively learns 'good' knowlcdge for prol)len) solving via interaction a.mong t,he aget1t.s withiit. esplicit, control mechanisms nor a global opt.imizat.ioii funct.ioii. To validate t,he effect,iveiiess of OCS, we have conducted intensive experiments on a real scale PCB design problem for electric appliances. 'i'lic. esperiment.al resuks have suggested that, (I ) OCS lias found feasible solutions wit,li t.he same quality of t.he ones by 1iuma.n espert.s; (2) t.1~ solutions are not. only locally opt.ima1, but also globally better than the ones by human experts wit,li regard t,o t,he total wiring length; a,nd (3) t.he soliitions are nmre preferable,t.Iian t,lie ones from the conventional C:omput.er Aided Design (CAD) syst.enls.