Naoya Takeishi
(武石 直也)

機械学習などをやっています。西スイス応用科学大学 機械学習グループ 研究員、理研AIP 客員研究員。

Naoya Takeishi is a researcher working at the Data Mining and Machine Learning Group of Haute école spécialisée de Suisse occidentale (HES-SO) Genève, Switzerland. He is a visiting scientist at Structured Learning Team of RIKEN Center for Advanced Intelligence Project (AIP), Japan. He is interested in effective integration of domain knowledge into statistical machine learning and also in data-driven analysis of dynamical systems, whose application ranges from scientific research to industrial machinery.

Research Interests

Knowledge-augmented machine learning.  Effective integration of domain-specific prior knowledge / inductive bias (e.g., mathematical models of physical laws, simulators, logical rules, and side information) into statistical machine learning. For example, we developed a deep dynamics model with prior knowledge of stable invariant sets such as limit cycles. We formulated a method for learning dynamical systems with side information. We proposed a regularized learning method for properly learning a physics-integrated (grey-box) VAEs, which are useful for robust and physically-interpretable generative modeling.

Data-driven analysis of dynamical systems.  Analysis of dynamical systems based on data-driven methods, such as dynamic mode decomposition (DMD) and its variants. We proposed, e.g., Bayesian DMD, DMD for random dynamics, DMD with neural net observables, nonnegative DMD, time-varying DMD, and discriminant DMD for labeled time-series. We introduced the use of multiple kernel learning for diffusion-map-based Koopman analysis.

Anomaly detection and interpretation.  Application of anomaly detection techniques based on machine learning to engineering systems, such as artificial satellites, vehicles, and power plants, as well as methodology for interpretation of anomaly detection. For example, we investigated the use of Shapley values for interpreting semi-supervised anomaly detection.

Visual SLAM in space.  Simultaneous estimation of the shape and motion of a target celestial body (e.g., asteroid) as well as the position and attitude of a spacecraft.


I no longer use old addresses (,; please use the latest ones above or my gmail.