Kohsuke Kubota

Applied Research Engineer

Sakana AI

自己紹介

私はSakana AIのApplied Research Engineerです。研究関心はマーケティングサイエンス、ベイズ統計学、因果推論、機械学習、深層学習、生成AIです。 慶應義塾大学にて経済学の博士号を取得し、星野崇宏教授のご指導をいただきました。 加えて、横浜市立大学データサイエンス研究科の客員研究員を務めています。

興味・関心
  • マーケティングサイエンス
  • ベイズ統計学
  • 因果推論
  • 機械学習
  • 深層学習
  • 生成AI
学歴

  • 経済学博士

    慶應義塾大学

  • 情報学修士

    京都大学

  • 工学士

    京都大学

Featured Publications
Off-Policy Evaluation

Off-Policy Evaluation and Learning for Survival Outcomes Under Censoring

In Proceedings of the 35th International Joint Conference on Artificial Intelligence (IJCAI) (Acceptance Rate = 13.6%, 713/5233)

Abstract Optimizing survival outcomes, such as patient survival or customer retention, is a critical objective in data-driven decision-making. Off-Policy Evaluation (OPE) provides a powerful framework for assessing such decision-making policies using logged data alone, without the need for costly or risky online experiments in high-stakes applications. However, typical estimators are not designed to handle right-censored survival outcomes, as they ignore unobserved survival times beyond the censoring time, leading to systematic underestimation of the true policy performance. To address this issue, we propose a novel framework for OPE and Off-Policy Learning (OPL) tailored for survival outcomes under censoring. Specifically, we introduce IPCW-IPS and IPCW-DR, which employ the Inverse Probability of Censoring Weighting technique to explicitly deal with censoring bias. We theoretically establish that our estimators are unbiased and that IPCW-DR achieves double robustness, ensuring consistency if either the propensity score or the outcome model is correct. Furthermore, we extend this framework to constrained OPL to optimize policy value under budget constraints. We demonstrate the effectiveness of our proposed methods through simulation studies and illustrate their practical impacts using public real-world data for both evaluation and learning tasks.

avatar
Kohsuke Kubota
•
Causal Inference

Causal Inference under Threshold Manipulation: Bayesian Mixture Modeling and Heterogeneous Treatment Effects

In Proceedings of the The 40th Annual AAAI Conference on Artificial Intelligence (AAAI) (Acceptance Rate = 17.6%, 4167/23680)

Abstract Many marketing applications, including credit card incentive programs, offer rewards to customers who exceed specific spending thresholds to encourage increased consumption. Quantifying the causal effect of these thresholds on customers is crucial for effective marketing strategy design. Although regression discontinuity design is a standard method for such causal inference tasks, its assumptions can be violated when customers, aware of the thresholds, strategically manipulate their spending to qualify for the rewards. To address this issue, we propose a novel framework for estimating the causal effect under threshold manipulation. The main idea is to model the observed spending distribution as a mixture of two distributions: one representing customers strategically affected by the threshold, and the other representing those unaffected. To fit the mixture model, we adopt a two-step Bayesian approach consisting of modeling non-bunching customers and fitting a mixture model to a sample around the threshold. We show posterior contraction of the resulting posterior distribution of the causal effect under large samples. Furthermore, we extend this framework to a hierarchical Bayesian setting to estimate heterogeneous causal effects across customer subgroups, allowing for stable inference even with small subgroup sample sizes. We demonstrate the effectiveness of our proposed methods through simulation studies and illustrate their practical implications using a real-world marketing dataset.

avatar
Kohsuke Kubota
•
Bayesian Estimation

Bayesian Time-Varying Meta-Analysis via Hierarchical Mean-Variance Random-effects Models

In Japanese Journal of Statistics and Data Science (Impact Factor = 1.0)

Abstract Meta-analysis is widely used to integrate results from multiple experiments to obtain generalized insights. Since meta-analysis datasets are often heteroscedastic due to varying subgroups and temporal heterogeneity arising from experiments conducted at different time points, the typical meta-analysis approach, which assumes homoscedasticity, fails to adequately address this heteroscedasticity among experiments. This paper proposes a new Bayesian estimation method that simultaneously shrinks estimates of the means and variances of experiments using a hierarchical Bayesian approach while accounting for time effects through a Gaussian process. This method connects experiments via the hierarchical framework, enabling “borrowing strength” between experiments to achieve high-precision estimates of each experiment’s mean. The method can flexibly capture potential time trends in datasets by modeling time effects with the Gaussian process. We demonstrate the effectiveness of the proposed method through simulation studies and illustrate its practical utility using a real marketing promotions dataset.

avatar
Kohsuke Kubota
•
Stock Recommendation

Content-based stock recommendation using smartphone data

In Journal of Information Processing, Specially Selected Paper (top 10%)

Abstract The number of investors holding risky assets in Japan is much lower than that in western countries even though it is an effective way for building investor assets. Although Japanese investment companies offer a service to invest in points through coalition loyalty programs instead of actual currency to address this situation, the problem still persists. One reason for this is that novice investors do not know in which stocks to invest. One possible solution is recommending stocks; however, we still face the cold-start problem because there is no transaction history for novice investors. In this study, we propose a novel content-based recommendation approach that utilizes touchpoint information, eg, payment and app usage data, on smartphones in daily life. This approach employs user-weighted recency, frequency, and monetary, called UW-RFM and a complementary module to comply with Japanese guidelines that prohibit presenting only a small number of companies and establishing a minimum number of companies to be presented. We conduct an online evaluation to validate the effectiveness of the proposed approach in an actual investment service. The evaluation results show that the proposed approach motivates users to invest more, ie, 0.352 more clicks on the recommendation area and 3,016 points (yen), than the baseline method that does not consider touchpoint information.

avatar
Kohsuke Kubota
•
Publications
Yuki Murakami, Takumi Hattori, Kohsuke Kubota (2026). Barycentric Fused Gromov-Wasserstein Balancing for Causal Inference under Multiple Treatments. arXiv preprint arXiv:2608.22024.
PDF 引用 DOI arXiv
Takumi Hattori, Kohsuke Kubota, Keiichi Ochiai (2026). Wald-Difference-in-Differences Estimation without Individual-level Treatment Data. In Journal of Information Processing.
PDF 引用 DOI
Yuki Murakami, Takumi Hattori, Kohsuke Kubota (2025). Multiple Treatments Causal Effects Estimation with Task Embeddings and Balanced Representation Learning. arXiv preprint arXiv:2511.09814.
PDF 引用 DOI arXiv
Kohsuke Kubota, Shonosuke Sugasawa (2025). Causal Inference under Threshold Manipulation: A Bayesian Mixture Approach. In KDD 2025 3rd Workshop on Causal Inference and Machine Learning in Practice.
PDF 引用
Yuki Murakami, Kohsuke Kubota, Takumi Hattori, Keiichi Ochiai (2025). Estimation of Single and Synergistic Treatment Effects under Multiple Treatments with Deep Neural Networks. In KDD 2025 3rd Workshop on Causal Inference and Machine Learning in Practice.
PDF 引用
Kohsuke Kubota, Keiichi Ochiai, Takahiro Hoshino (2025). Causal effect of lottery promotions on post-win payments: Evidence from a large field experiment. In Innovative Marketing (Impact Factor = 1.2, Acceptance Rate = 33%).
PDF 引用 DOI
Takumi Hattori, Kohsuke Kubota, Keiichi Ochiai (2025). Wald-Differences-in-Differences Estimation without Individual-Level Treatment Data. In AAAI’25 Workshop on Artificial Intelligence with Causal Techniques.
PDF 引用
Hisao Katsumi, Kohsuke Kubota, Wataru Yamada, Keiichi Ochiai (2023). Stay Ahead of the Competition: An Approach for Churn Prediction by Leveraging Competitive Service App Usage Logs. In UbiComp/ISWC ‘23 Adjunct Proceedings of the 2023 ACM International Joint Conference on Pervasive and Ubiquitous Computing & the 2023 ACM International Symposium on Wearable Computing.
引用 DOI
Kohsuke Kubota, Keiichi Ochiai (2020). Time-aware GCN: Representation Learning for Mobile App Usage Time-series Data. In KDD 2020 The Second International Workshop on Deep Learning on Graphs:Methods and Applications.
PDF 引用
Academic Service

Conference Program Committee

Journal Reviewer