GENET and its extended model EGENET are artificial neural networksto efficiently solve finite constraint satisfaction problems such as thecar-sequencing problems. Both models use the min-conflict heuristic toupdate every finite-domain variable for finding local minima, and thenapply heuristic learning rule(s) to escape the local minima notrepresenting solution(s). Since continuous and finite domains arecompletely different, researchers seldom considered to apply the EGENETapproach to solve continuous constrained optimization problems. Weconsider an interesting proposal to modify the original EGENET modelwith the minimal effort for solving continuous constrained optimizationproblems. Our proposal immediately opens up new directions for studyingmany possible ways to approximate continuous domains using modifiedfinite-domain solvers. Moreover, the preliminary benchmarks of ourprototypes on some graph layout problems as practical examplesdemonstrated some advantages of our proposal which prompts for furtherinvestigation
展开▼
