Causal discovery in the presence of hidden confounding variables for data with heterogeneity

• Project/Area Number: 16K00045
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Statistical science
• Research Institution: Shiga University
• Principal Investigator: Shimizu Shohei 滋賀大学, データサイエンス学部, 教授 (10509871)
• Project Period (FY): 2016-04-01 – 2020-03-31
• Project Status: Completed (Fiscal Year 2019)
• Budget Amount: ¥4,680,000 (Direct Cost: ¥3,600,000、Indirect Cost: ¥1,080,000); Fiscal Year 2019: ¥1,040,000 (Direct Cost: ¥800,000、Indirect Cost: ¥240,000); Fiscal Year 2018: ¥780,000 (Direct Cost: ¥600,000、Indirect Cost: ¥180,000); Fiscal Year 2017: ¥910,000 (Direct Cost: ¥700,000、Indirect Cost: ¥210,000); Fiscal Year 2016: ¥1,950,000 (Direct Cost: ¥1,500,000、Indirect Cost: ¥450,000)
• Keywords: 因果探索 / 因果構造 / 観察データ / 未観測共通原因 / 異質性 / 統計的因果推論 / 構造的因果モデル / 因果構造探索

Outline of Final Research Achievements

LiNGAM model handles only continuous variables. To represent heterogeneity, we tried to extend the LiNGAM model so that it can handle discrete variables. We developed a model assuming that the relationship between discrete variables and continuous variables is a non-cyclic directed graph. We also considered combining a causal model with a machine learning model that can handle discrete variables. To deal with unobserved common causes, we extended instrumental variable methods by making use of non-Gaussianity and independence. In addition, a method to infer where the unobserved common cause is likely to be is proposed within the framework of the LiNGAM model.

Academic Significance and Societal Importance of the Research Achievements

LiNGAMモデルは因果探索の標準的な方法の一つとして注目を集めているが, 離散変数が混在する状況を扱えるようにすることでさらに応用範囲を広げることができた。また機械学習モデルと因果モデルを組み合わせたモデルについては,制御への応用が期待される。操作変数は広く用いられているが，非ガウス性と独立性を利用した操作変数法については,従来よりも多くの情報を抽出することができることがわかった。未観測共通原因がどこにありそうかを推測する方法については, 条件付き独立性を用いる因果関係推測法の枠組みでは, そのような方法が提案されているが, LiNGAMモデルの枠組みでは対応する方法がなかった。

Research Products

• [Int'l Joint Research] Hunan University of Commerce(中国)
• [Int'l Joint Research] The University of Auckland(ニュージーランド)
• [Int'l Joint Research] Shanghai University(中国)
• [Int'l Joint Research] Max Planck institute(ドイツ)
• [Int'l Joint Research] University College London(United Kingdom)
• [Journal Article] RCD: Repetitive causal discovery of linear non-Gaussian acyclic models with latent confounders (2020)
  • Author(s): T. N. Maeda, S Shimizu
  • Journal Title: Proc. 23rd International Conference on Artificial Intelligence and Statistics (AISTATS2020) Volume: 1 Pages: 1-9
  • Peer Reviewed / Open Access
• [Journal Article] Multi-Modality Behavioral Influence Analysis for Personalized Recommendations in Health Social Media Environment (2019)
  • Author(s): Zhou Xiaokang、Liang Wei、Wang Kevin I-Kai、Shimizu Shohei
  • Journal Title: IEEE Transactions on Computational Social Systems Volume: 6 Issue: 5 Pages: 888-897
  • DOI: 10.1109/tcss.2019.2918285
  • Peer Reviewed
• [Journal Article] Analysis of cause-effect inference by comparing regression errors (2019)
  • Author(s): Blobaum Patrick、Janzing Dominik、Washio Takashi、Shimizu Shohei、Scholkopf Bernhard
  • Journal Title: PeerJ Computer Science Volume: 5 Pages: e169-e169
  • DOI: 10.7717/peerj-cs.169
  • Peer Reviewed / Open Access / Int'l Joint Research
• [Journal Article] Personalization Recommendation Algorithm Based on Trust Correlation Degree and Matrix Factorization (2019)
  • Author(s): Li Weimin、Zhou Xiaokang、Shimizu Shohei、Xin Mingjun、Jiang Jiulei、Gao Honghao、Jin Qun
  • Journal Title: IEEE Access Volume: 7 Pages: 45451-45459
  • DOI: 10.1109/access.2018.2885084
  • Peer Reviewed / Int'l Joint Research
• [Journal Article] A Novel Personalized Recommendation Algorithm Based on Trust Relevancy Degree (2018)
  • Author(s): Li Weimin、Zhu Heng、Zhou Xiaokang、Shimizu Shohei、Xin Mingjun、Jin Qun
  • Journal Title: Proc. DASC/PiCom/DataCom/CyberSciTec2018 Volume: 1 Pages: 418-422
  • DOI: 10.1109/dasc/picom/datacom/cyberscitec.2018.00084
  • Peer Reviewed / Int'l Joint Research
• [Journal Article] Cause-Effect Inference by Comparing Regression Errors (2018)
  • Author(s): Patrick Bloebaum, Dominik Janzing, Takashi Washio, Shohei Shimizu, Bernhard Schoelkopf
  • Journal Title: Proceedings of the Twenty-First International Conference on Artificial Intelligence and Statistics (AISTATS2018), PMLR Volume: 84 Pages: 900-909
  • Peer Reviewed / Open Access / Int'l Joint Research
• [Journal Article] Non-Gaussian Methods for Causal Structure Learning (2018)
  • Author(s): Shimizu Shohei
  • Journal Title: Prevention Science Volume: 20 Issue: 3 Pages: 431-441
  • DOI: 10.1007/s11121-018-0901-x
  • Peer Reviewed / Open Access
• [Journal Article] Learning instrumental variables with structural and non-Gaussianity assumptions (2017)
  • Author(s): Ricard Silva, Shohei Shimizu
  • Journal Title: Journal of Machine Learning Research Volume: 18 Pages: 1-49
  • Peer Reviewed / Open Access / Int'l Joint Research
• [Journal Article] Error asymmetry in causal and anticausal regression (2017)
  • Author(s): Blobaum Patrick、Washio Takashi、Shimizu Shohei
  • Journal Title: Behaviormetrika Volume: 44 Issue: 2 Pages: 491-512
  • DOI: 10.1007/s41237-017-0022-z
  • Peer Reviewed / Int'l Joint Research
• [Journal Article] A novel principle for causal inference in data with small error variance (2017)
  • Author(s): Blobaum Patrick、Shimizu Shohei、Washio Takashi
  • Journal Title: n Proc. 25 th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning (ESANN2017), Volume: 1 Pages: 347-352
  • Peer Reviewed
• [Journal Article] Estimation of interventional effects of features on prediction (2017)
  • Author(s): Blobaum Patrick、Shimizu Shohei
  • Journal Title: Proc. 2017 IEEE Machine Learning for Signal Processing Workshop (MLSP2017) Volume: 1 Pages: 1-6
  • Peer Reviewed
• [Journal Article] Non-Gaussian structural equation models for causal discovery (2016)
  • Author(s): S. Shimizu
  • Journal Title: Statistics and Causality: Methods for Applied Empirical Research Volume: - Pages: 153-184
  • Peer Reviewed
• [Presentation] Causal discovery based on non-Gaussianity of data and its applications (2019)
  • Author(s): Shohei SHIMIZU
  • Organizer: 日本行動計量学会 第47回大会
  • Invited
• [Presentation] 統計的因果探索に基づく遺伝子制御ネットワークの推定 (2019)
  • Author(s): 井元佑介, 平岡裕章，清水昌平，前田高志ニコラス，小島洋児，斎藤通紀
  • Organizer: 応用数学合同研究集会2019
• [Presentation] Causal discovery, prediction, and control (2018)
  • Author(s): Shohei Shimizu
  • Organizer: Causal Modeling and Machine Learning (CaMaL) Workshop, Guangzhou, China.
  • Int'l Joint Research / Invited
• [Presentation] Causal discovery, prediction mechanisms, and control (2018)
  • Author(s): Shohei Shimizu
  • Organizer: he 5th meeting of the Institute of Mathematical Statistics (IMS) meeting series, the IMS Asia Pacific Rim Meeting (IMS-APRM), Singapore
  • Int'l Joint Research / Invited
• [Presentation] 因果探索、予測、そして制御 (2018)
  • Author(s): 清水昌平
  • Organizer: 2018年度 統計関連学会連合大会, 東京. 応用統計学会企画セッション: 「統計的因果推論―基本的なアイデアから最近の発展まで―」
  • Invited
• [Presentation] 因果構造探索の基本 (2017)
  • Author(s): 清水昌平
  • Organizer: 研究集会: 因果推論の基礎
  • Place of Presentation: 統計数理研究所 (東京)
  • Year and Date: 2017-02-17
  • Invited
• [Presentation] 統計的因果推論への招待 - 因果構造探索を中心に - (2017)
  • Author(s): 清水 昌平
  • Organizer: システム制御情報学会・計測自動制御学会 チュートリアル講座2017
  • Invited
• [Presentation] Causal discovery and prediction mechanisms (2017)
  • Author(s): Shohei Shimizu
  • Organizer: France/Japan Machine Learning Workshop
  • Int'l Joint Research / Invited
• [Presentation] 因果探索への招待 (2017)
  • Author(s): 清水 昌平
  • Organizer: 電子情報通信学会IA(インターネットアーキテクチャ)/IN(情報ネットワーク)併催研究会
  • Invited
• [Presentation] 因果探索入門 (2017)
  • Author(s): 清水 昌平
  • Organizer: 日本行動計量学会 第20回春の合宿セミナー
  • Invited
• [Presentation] 関係流動性と消費者自民族中心主義の因果構造分析～非ガウス性を使った因果推論 (2016)
  • Author(s): 芳賀麻誉美, 清水昌平
  • Organizer: 日本マーケティング・サイエンス学会 第100回研究大会
  • Place of Presentation: ホテル阪急エキスポパーク＋大阪大学中之島センター (大阪)
  • Year and Date: 2016-12-27
• [Presentation] A non-Gaussian model for causal discovery in the presence of hidden common causes (2016)
  • Author(s): S. Shimizu
  • Organizer: A non-Gaussian model for causal discovery in the presence of hidden common causes
  • Place of Presentation: Munich (Germany)
  • Year and Date: 2016-05-23
  • Int'l Joint Research / Invited
• [Presentation] A non-Gaussian approach for causal structure learning in the presence of hidden common causes (2016)
  • Author(s): S. Shimizu
  • Organizer: CRM Workshop: Statistical Causal Inference and its Applications to Genetics
  • Place of Presentation: Montreal (Canada)
  • Int'l Joint Research / Invited
• [Book] 統計的因果探索 (2017)
  • Author(s): 清水 昌平
  • Total Pages: 192
  • Publisher: 講談社
  • ISBN: 9784061529250
• [Book] 確率的グラフィ カルモデル (2016)
  • Author(s): 黒木学， 清水昌平， 湊真一， 石畠正和， 樺島祥介， 田中和之， 本村陽一， 玉田嘉紀， 鈴木譲， 植野真臣
  • Total Pages: 292
  • Publisher: 共立出版
• [Remarks] https://sites.google.com/site/sshimizu06/
• [Remarks] 清水昌平
  • URL: https://sites.google.com/site/sshimizu06/indexj
• [Remarks]
  • URL: https://sites.google.com/site/sshimizu06/