Analysis of models of data from the perspectives of theories of scientific representation

2022 Fiscal Year Final Research Report

• Project/Area Number: 18K12178
• Research Category: Grant-in-Aid for Early-Career Scientists
• Allocation Type: Multi-year Fund
• Review Section: Basic Section 01010:Philosophy and ethics-related
• Research Institution: The Graduate University for Advanced Studies
• Principal Investigator: Onishi Yukinori 総合研究大学院大学, 統合進化科学研究センター, 講師 (50793155)
• Project Period (FY): 2018-04-01 – 2023-03-31
• Keywords: データのモデル / 科学的表象 / 科学的実在論論争 / データ同化 / 深層学習 / 科学的理解

Outline of Final Research Achievements

In this study, I focused on so-called models of data, which are located at the interface between data and the world, and analyzed them from the perspectives of scientific representation and the scientific realism debate. In the course of research, however, I switched my attention to peculiar modes of creating data models with emerging technologies, i.e., data assimilation and deep learning. Regarding data assimilation, through participation in the Winter School at RIKEN and holding a symposium with data assimilation researchers, I collected basic knowledge and information on research practices with data assimilation, and obtained insights for future studies. For the analysis of deep learning (DL), I concluded that, while it has little impact on the structure of the traditional realist-anti-realist debate, various interpretation methods used in DL studies could provide interesting insights for the relationship between scientific understanding and realism.

Free Research Field

科学哲学

Academic Significance and Societal Importance of the Research Achievements

データ同化に関する科学哲学的分析は、海外では徐々に行われてきているが、国内の科学哲学者のあいだではあまり知られておらず、また深層学習技術と実在論との関係も、一部の例外を除いて国内外でほとんど議論が行われていない。これら新興技術の哲学的分析は、非常に時宜を得たものであるといえる。