2018 Fiscal Year Final Research Report

Distributed/diversified/case-based evolutionary intelligence and its evaluation of applicability to industry and medical apparatus

Research Project

PDF

Project/Area Number	15K00349
Research Category	Grant-in-Aid for Scientific Research (C)
Allocation Type	Multi-year Fund
Section	一般
Research Field	Soft computing
Research Institution	Meiji University (2018) Tokyo Denki University (2015-2017)
Principal Investigator	Tsuruta Setsuo 明治大学, 研究・知財戦略機構, 研究推進員(客員研究員) (00366395)
Co-Investigator(Kenkyū-buntansha)	小橋昌司兵庫県立大学, 工学研究科, 教授 (00332966) 八槇博史東京電機大学, システムデザイン工学部, 教授 (10322166) 櫻井義尚明治大学, 総合数理学部, 専任准教授 (30408653) 川辺孝東京電機大学, システムデザイン工学部, 教授 (40339081)
Project Period (FY)	2015-04-01 – 2019-03-31
Keywords	遺伝的アルゴリズム / 局所解法 / 産業応用 / 事例
Outline of Final Research Achievements	Case-based evolutionary intelligence was developed. It is called CBGALO: Case Based (CB) Genetic Algorithm (GA) integrated with Local Optimizer (LO). Local optimizers were such as Support Vector Machine (SVM) or heuristics. The generic applicability to industry and medical apparatus was evaluated. After evaluation in delivery route scheduling, it was applied to class imbalanced problems such as prediction of solar flare causing serious problems on the human. CBGALO was elaborated to be case based automatically restart-able GA searching both learning features and learning data. Even Deep Learning (DL) cannot search the latter. Further, diversity was maintained by hamming distance to elite individuals. The precision reached over 90%, surpassing the conventional world top level. For automated knee bone segmentation in MR images, CBGALO was applied to find an appropriate cell set by evolutionary multiple classifiers dependent on location. Precision rapidly reached over 65%.
Free Research Field	情報工学、人工知能
Academic Significance and Societal Importance of the Research Achievements	１）事例や学習データの進化活用により、災害予測などクラス不均衡問題の解精度でも学術的に世界最高水準を凌駕した。２）専門家が作成／修正した部分を事例として破壊せず利用することで、社会的条件が絡む実用の問題も解決可能とした。３）物流・防災を含む産業、医工学の複雑な最適化問題の解決が可能となり、進化的最適化技術利用の促進につながる。４）本提案による最適化は、1兆円以上かつ増加傾向にある物流市場の配送コスト低減に寄与する。５）鉄道の５倍以上の配送用トラックの温室効果ガスＣＯ２排出の削減にも貢献する。６）医工・防災産業を含めると10兆円以上の市場となり、経済効果・社会的意義は更に高い。