吉田直生

機械学習の研究で博士課程に在籍中です。
深層学習の理論解析などを行っています。

研究分野：深層学習理論、確率的最適化理論

所属：東京大学総合文化研究科広域科学専攻

Google Scholar: link

LinkedIn: link

論文

Naoki Yoshida*, Satoshi Hayakawa*, Yuhta Takida, Toshimitsu Uesaka, Hiromi Wakaki, Yuki Mitsufuji, "Theoretical Refinement of CLIP by Utilizing Linear Structure of Optimal Similarity", Transactions on Machine Learning Research, 2026. paper

Naoki Yoshida, Shogo Nakakita, Masaaki Imaizumi, "Effect of Random Learning Rate: Theoretical Analysis of SGD Dynamics in Non-Convex Optimization via Stationary Distribution", Transactions on Machine Learning Research, 2025. paper movie code
Naoki Yoshida, Sumio Watanabe, "Upper Bound of Real Log Canonical Threshold of Tensor Decomposition and its Application to Bayesian Inference", International Symposium on Stochastic Systems Theory and Its Applications, 2023. paper

Naoki Yoshida, Isao Ishikawa, Masaaki Imaizumi, "Zero Generalization Error Theorem for Random Interpolators via Algebraic Geometry", arxiv preprint arXiv:2512.06347, 2025. paper

Page updated

Google Sites

Report abuse