Automatic Generation of Graph-structural Programs by Using Swarm Intelligence of Ants

2014 Fiscal Year Final Research Report

• Project/Area Number: 25730149
• Research Category: Grant-in-Aid for Young Scientists (B)
• Allocation Type: Multi-year Fund
• Research Field: Soft computing
• Research Institution: Hiroshima City University
• Principal Investigator: HARA Akira 広島市立大学, 情報科学研究科, 准教授 (70347615)
• Project Period (FY): 2013-04-01 – 2015-03-31
• Keywords: 群知能 / アリコロニー最適化法 / 遺伝的プログラミング

Outline of Final Research Achievements

The aim of this research is to develop an automatic generation method of graph-structural programs by using swarm intelligence. We have already proposed Cartesian Ant Programming (CAP) as an automatic programming method. But, CAP search tends to fall into local optima because assignments of functional symbols (program elements) to respective nodes are fixed and inflexible. In my 2013's research, we have proposed an improved CAP method, CAP with adaptive node replacements. The assignments of functional symbols changes adaptively based on the state of search in this method. In my 2014's research, for improving the search performance, we have introduced two new mechanisms into CAP, optimization of function assignments by pheromone communication, and limitation or release of the use of a part of nodes. These methods showed better performance than conventional CAP.

Free Research Field

進化計算