教員一覧 List of faculty members

SHIMAUCHI Hirokazu 准教授

SHIMAUCHI, Hirokazu Associate Professor

Research Projects

We are working on developing and applying machine learning methods to uncover hidden patterns in data. In particular, we aim to capture the geometric structure behind the data and develop efficient machine learning methods that function even with limited computational resources. In recent years, we have been working on representation learning and outlier detection based on pseudoconformal mapping theory, new activation and representation transformation methods in deep learning, and feature representation learning based on non-commutative quantum dynamics utilizing quantum computers. We are also conducting applied research of machine learning in fields such as social sciences and fluid dynamics analysis, in collaboration with researchers from other fields.

The appeal of research

I find it appealing that we can freely think about algorithms related to how machines learn from data, using new ideas and approaches. We aim for research that "turns 0 into 1," without being bound by existing frameworks.

Achievements

• Best Paper Candidates (Top 3), IEEE/ACM International Conference on Big Data Computing, Applications and Technologies, 2019
• Best Paper Candidates, 15th International Joint Conference on Computational Intelligence: Neural Computation Theory and Applications, 2023
• Received the Aomori Prefecture Industrial Technology Education Promotion Association Encouragement Award, 2024
• Best Paper Candidates, Geometry-Aware Stochastic Structured Channel Mixing for Deep Neural Networks, 6th International Conference on Image Processing and Vision Engineering, 2026

Major publications and papers

• H. Shimauchi, Noncommutative Quantum Dynamics for Feature Representation Learning, IEEE Access, 14, 53529-53571, 2026 (peer-reviewed).
• H. Shimauchi, Geometry-Aware Stochastic Structured Channel Mixing for Deep Neural Networks, Accepted for publication in the Proceedings of 6th International Conference on Image Processing and Vision Engineering, Communications in Computer and Information Science, Springer, 2026 (peer-reviewed).
• SHIMAUCHI, Hirokazu, Numerical Construction Method of Pseudoconformal Mapping and its Application to Machine Learning, IEICE Fundamentals Review, 19(2), 97-104, 2025 (Invited).
• H. Shimauchi, Quasiconformal Extension-Based Unsupervised Representation Learning and Application to Semi-supervised Outlier detection, Computational Intelligence, 187–210, 2025 (peer-reviewed).
• H. Shimauchi, Unsupervised Representation Learning by Quasiconformal Extension, In Proceedings of the 15th International Joint Conference on Computational Intelligence, 1, 440-449, 2023 (peer-reviewed).
• H. Shimauchi, An Activation Function with Probabilistic Beltrami Coefficient for Deep Learning, In Proceedings of the 14th International Conference on Agents and Artificial Intelligence, 3, 613-620, 2022 (peer-reviewed).
• H. Shimauchi, Improving Supervised Outlier Detection by Unsupervised Representation Learning and Generative Adversarial Networks, In Proceedings of the 4th International Conference on Information Science and Systems, 22-27, 2021 (peer-reviewed).
• S. Kato, T. Nakanishi, B. Ahsan, H. Shimauchi, Time-series topic analysis using singular spectrum transformation for detecting political business cycles, Journal of Cloud Computing, 10, 21, 1-16, 2021 (peer-reviewed).
• S. Kato, T. Nakanishi, H. Shimauchi, B. Ahsan, Topic Variation Detection Method for Detecting Political Business Cycles, In Proceedings of the 6th IEEE/ACM International Conference on Big Data Computing, Applications and Technologies, 85-93, 2019 (peer-reviewed).
• Tohoku University Institute for Interdisciplinary Science Frontiers, "Encyclopedia of Science" Editorial Committee, Encyclopedia of Science Vol. 1: Interdisciplinary Frontiers Challenged by Young Researchers (Contributed section: Chapter 3), Tohoku University Press, 2019.
  
  (Collaborative research with research students and graduate school students from the laboratory)
• Y. Kanda, H. Shimauchi, Y. Katori, Knowledge Distillation From TSMixer to Echo StateNetwork for Multivariate Time-Series Forecasting, IEICE GlobalNet Workshop 2026, 2026.
• Sohei Ima, Hirokazu Shimauchi, A lightweight deep learning model using depth-separated convolution and knowledge distillation for unsupervised anomaly detection suitable for edge environments, IEICE General Conference (Sensor Networks and Mobile Intelligence), 2026.
• Akira Tada, Hirokazu Shimauchi, Reika Nomura, Kenta Tozato, Shuji Moriguchi, Kenjiro Terada, Shinsuke Takase, "Physics-Informed Neural Network Based on Latent Representation Learning of Observational Information for Immediate Prediction of Tsunami Wave Fields," Proceedings of the 29th Symposium on Applied Mechanics (Section 2, 5), 2026.
• Akira Tada, Hirokazu Shimauchi, Reika Nomura, Shuji Moriguchi, Kenjiro Terada, Shinsuke Takase, Kenta Tozato, Latent Representation Learning of Offshore Tsunami Observation Data and Real-Time Tsunami Prediction using a Physics-Informed Neural Network, IEICE General Conference (Artificial Intelligence and Knowledge Processing), 2026.