Short Biography: Koshi Watanabe is a 2nd-year Ph.D student in Information Science and Technology at Hokkaido University. He earned his B.S. in Information Science and Technology at Hokkaido University (2022) and also earned his M.S. degree in Information Science and Technology at Hokkaido University (2024), advised by Prof. Takahiro Ogawa.
his research interest lies in statistical machine learning (e.g., Gaussian process) and unsupervised learning (e.g., dimensionality reduction or sparse covariance estimation). He also engaged the application of machine learning algorithms to the damage assessment of infrastructure images.
koshi [at] lmd.ist.hokudai.ac.jp
Biography
- 2024/04 ~ Ph. D student in Information Science, Hokkaido University.
- 2022/04 ~ 2024/03 Master of in Information Science, Hokkaido University.
- 2018/04 ~ 2022/03 Bachelor of Engineering, Hokkaido University.
Publications
Preprint
- 〇Koshi Watanabe, Keisuke Maeda, Takahiro Ogawa and Miki Haseyama, “StarMAP: Global neighbor embedding for faithful data visualization,” arXiv preprint arXiv:2502.03776, 2025. [paper] [code (google drive)]
Journal
- 〇Koshi Watanabe, Keisuke Maeda, Takahiro Ogawa and Miki Haseyama, “SpectralMAP: Approximating data manifold with spectral decomposition,” IEEE Access, vol. 11, pp. 31530-31540, 2023. (2022 IF=3.9) [paper]
- 〇Koshi Watanabe, Keisuke Maeda, Takahiro Ogawa and Miki Haseyama, “Gaussian process decoder with spectral mixtures and locally estimated manifold for data visualization,” Applied Sciences, vol. 13, no. 8018, pp. 1-16, 2023. (2022 IF=2.7) [paper]
International Conference
- 〇Koshi Watanabe, Keisuke Maeda, Takahiro Ogawa and Miki Haseyama, “Hyperboloid GPLVM for discovering continuous hierarchies via nonparametric estimation,” International Conference on Artificial Intelligence and Statistics (AISTATS), 2025 (583/1861=31.3%). [paper] [code]
- 〇Koshi Watanabe, Keisuke Maeda, Takahiro Ogawa and Miki Haseyama, “Estimating graph topology with smooth latent signals via Gaussian process,” International Workshop on the New Frontiers in Convergence Science and Technology: The 26th HU-SNU Joint Symposium Satellite Session, 2023.
- 〇Koshi Watanabe, Keisuke Maeda, Takahiro Ogawa and Miki Haseyama, “Learning graph Laplacian from intrinsic patterns via Gaussian process,” IEEE International Conference on Acoustics Speech and Signal Processing (ICASSP), 2023. [paper]
- 〇Koshi Watanabe, Keisuke Maeda, Takahiro Ogawa and Miki Haseyama, “Summarizing data structures with Gaussian process and robust neighborhood preservation,” European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML PKDD), pp.153-173, 2022 (242/932=26%). [paper]
- 〇Koshi Watanabe, Keisuke Maeda, Takahiro Ogawa and Miki Haseyama, “Distributed label dequantized Gaussian process latent variable model for multi-view data integration,” IEEE International Conference on Acoustics Speech and Signal Processing (ICASSP), pp.4643-4647, 2022. [paper]
- 〇Koshi Watanabe, Keisuke Maeda, Takahiro Ogawa and Miki Haseyama, “Movie rating estimation based on weakly supervised multi-modal latent variable model,” \IEEE Global Conference on Consumer Electronics (GCCE), pp.221–222, 2021. [paper]
Funding
- 2024.4-2027.3 JSPS Research Fellow DC1 (7,200,000JPY)
- 2023.4-2024.3 NITORI International Scholarship Foundation’s Scholarship for Future IT Hu-
man Resources (960,000JPY)