Learning of formal graph systems and its application to graph mining

2016 Fiscal Year Final Research Report

• Project/Area Number: 26280087
• Research Category: Grant-in-Aid for Scientific Research (B)
• Allocation Type: Partial Multi-year Fund
• Section: 一般
• Research Field: Intelligent informatics
• Research Institution: Kyushu International University
• Principal Investigator: Shoudai Takayoshi 九州国際大学, 現代ビジネス学部, 教授 (50226304)
• Co-Investigator(Kenkyū-buntansha): 内田 智之 広島市立大学, 情報科学研究科, 准教授 (70264934)
• Project Period (FY): 2014-04-01 – 2017-03-31
• Keywords: グラフ文法 / グラフパターン / 形式体系 / 計算論的学習 / 機械学習 / 帰納推論 / グラフマイニング / グラフアルゴリズム

Outline of Final Research Achievements

Formal graph system (FGS) is a logic program that deals with term graphs instead of the terms of first-order predicate logic. In this research, we have studied the polynomial-time learnability of graph patterns and formal graph systems in the frameworks of the PAC-learning, query-learning, and inductive inference models. In particular, we introduced the hierarchy of FGS that is defined by some numerical graph invariants and the parameters of logic programs. We showed that there exist formal graph systems in the hierarchy that are learnable in the PAC framework. Furthermore, we described how distributional learning techniques are applied to formal graph system (FGS) languages. We showed that the regular FGS languages of bounded degree with the 1-finite context property (1-FCP) and bounded treewidth property are learned from positive data and membership queries.

Free Research Field

計算機科学