SOTA MORIYAMA

Research Interest

My work focuses on building differentiable combinatorial solvers—primarily SAT and MaxSAT—by leveraging Deep Learning. By integrating these learnable, differentiable solvers directly into neuro-symbolic pipelines, I aim to create frameworks where symbolic reasoning and neural learning are seamlessly combined, leading to more robust and trustworthy AI systems.

Grants/Fellowship

• 2026.04 - 2028.03

  JSPS Research Fellowship for Young Scientists (DC2) Trustworthy Deep Models with Graph Learning Techniques Based on Logical Constraints

• 2025.10 - 2025.11

  SOKENDAI Student Dispatch Program Research visit to Professor Thomas Eiter's lab at TU Wien

• 2025.04 - 2026.03

  SOKENDAI Special Researcher (JST Spring) Neuro-Symbolic AI from Combination of Maximum Satisfiability and Graph Neural Networks

Award

• 2025.03

  Inose Student Encouragement Award, NII

• 2023.12

  ROAD-R Challenge for NeurIPS 2023 (Task 2): 1st Place Award

• 2023.12

  ROAD-R Challenge for NeurIPS 2023 (Task 1): 3rd Place Award

• 2023.03

  World AI Competition YAMAGUCHI (2022): 3rd Place Award