Research Interest

I explain what Kitaoka has primarily researched (as of Octorber 11, 2025).

Inverse optimization for Mixed-integer programming

• We aim to learn objective functions and constraints of MILP such that the given data is the optimal solution.
• Detail (Japanese)

Counterfactual explanation, Algorithmic recourse

• We research counterfactual explanations, i.e., AIs that explain “how to modify the current input to obtain the desired output”.
• Detail (Japanese)

Learning of averaged data

• Data aggregation techniques, including data averaging, are useful for data summarization, compression, reducing annotation costs, improving data quality, and protecting privacy.
• We are researching methods for machine learning from averaged data.
• Detail (Japanese)

Algorithm of approximating minimal geodesics

• I was conducting research on estimating errors in length minimization through energy minimization of curves in real coordinate space using finite difference methods and numerical integration.
• Detail (Japanese)

Invariants of Parabolic Geometry with BGG Complexes

• I was particularly interested in the relationship between geometric invariants of the de Rham complex and the (curved) Bernstein-Gelfand-Gelfand complex on contact geometry or parabolic geometry.
• Datial (Japanese)