2022 Fiscal Year Final Research Report
Causal discovery from data in the presence of unobserved common causes
Project/Area Number |
20K19872
|
Research Category |
Grant-in-Aid for Early-Career Scientists
|
Allocation Type | Multi-year Fund |
Review Section |
Basic Section 61030:Intelligent informatics-related
|
Research Institution | Tokyo Denki University (2022) Institute of Physical and Chemical Research (2020-2021) |
Principal Investigator |
|
Project Period (FY) |
2020-04-01 – 2023-03-31
|
Keywords | 統計的因果探索 / 因果推論 / 未観測変数 |
Outline of Final Research Achievements |
In this study, causal discovery methods have been developed in the presence of unobserved variables. Statistical causal discovery is the inference of causal relationships between variables based on certain assumptions about the observed data and the processes by which they are generated. Previous studies have assumed the absence of unobserved variables and could only be applied to limited types of data. In the present study, such assumptions were removed, allowing statistical causal discovery methods to be carried out for a wide range of data. In particular, this study was able to propose a method that can be applied to both cases where the causal function is linear and non-linear.
|
Free Research Field |
知能情報学
|
Academic Significance and Societal Importance of the Research Achievements |
これまでの統計的因果探索は未観測変数の存在を認めない仮定をおいており、応用可能なデータの範囲は限られたものであった。統計的因果探索は経済・生物・社会など、さまざまな領域においてすでに多方面での応用がなされており、すでに多くの成果が得られている。このため、未観測変数を持つデータへの応用が望まれていた。本研究では、この仮定を取り除くことができたため、これまで以上に統計的因果探索の応用範囲を広げることができた。因果とは、これまで経験したことのない介入を施した際に生じる結果であり、この推論ができる領域が増えたため、社会政策などさまざまな分野へ応用が可能である。
|