2008 | OriginalPaper | Chapter
The Performance of a Selection Architecture for Genetic Programming
Activate our intelligent search to find suitable subject content or patents.
Select sections of text to find matching patents with Artificial Intelligence. powered by
Select sections of text to find additional relevant content using AI-assisted search. powered by
Hierarchical decomposition and reuse techniques are seen as making a vital contribution to the scalability of genetic programming systems. Existing techniques either try to identify and encapsulate useful code fragments as they evolve, or else they rely on intelligent prior deconstruction of the problem at hand. The alternative we propose is to base decomposition on a partitioning of the input test cases into subsets or ranges. To effect this, the program architecture of individuals is such that each subset is dealt with in an independently evolved branch, rooted at a selection node handling branch activation. Experimentation reveals that performance of systems employing this architecture is significantly better than that of more conventional systems.