Shirakawa Laboratory

Our papers regarding search space design for NAS and open domain generalization have been accepted to International Joint Conference on Neural Networks (IJCNN 2024) (Part of WCCI 2024) held in Yokohama.

Our paper has been accepted to Pacific-Asia Conference on Knowledge Discovery and Data Mining (PAKDD 2024). This paper introduces the feature selection mechanism into Neural Additive Models and Neural Basis Models, which are interpretable machine learning models, and enables them to apply high-dimensional datasets.

Members’ Page has been updated. Now, our laboratory has 8 doctoral course students, 14 master’s course students, and 6 undergraduate students for graduation research.

Our paper regarding the conversion of tabular data to image data has been accepted to Knowledge-Based Systems.

• Takuya Matsuda, Kento Uchida, Shota Saito, and Shinichi Shirakawa: HACNet: End-to-end learning of interpretable table-to-image converter and convolutional neural network, Knowledge-Based Systems, Vol. 284, 111293, Jan. 2024. [DOI]

Our paper regarding the CMA-ES with Margin for mixed-integer black-box optimization problems has been accepted to ACM Transactions on Evolutionary Learning and Optimization. This paper is joint work with M. Nomura at CyberAgent, Inc.

• Ryoki Hamano, Shota Saito, Masahiro Nomura, and Shinichi Shirakawa: Marginal Probability-Based Integer Handling for CMA-ES Tackling Single-and Multi-Objective Mixed-Integer Black-Box Optimization, ACM Transactions on Evolutionary Learning and Optimization. [DOI] [arXiv]

Our paper regarding the automatic termination for neural architecture search has been accepted to Neural Networks. This work is collaborative research with Prof. Hino (Institute of Statistical Mathematics), etc.

• Kotaro Sakamoto, Hideaki Ishibashi, Rei Sato, Shinichi Shirakawa, Youhei Akimoto, and Hideitsu Hino: ATNAS: Automatic Termination for Neural Architecture Search, Neural Networks, Vol. 166, pp. 446-458, Sep. 2023. [DOI]

Yuta Suzuki, a master’s course student, presented about applying machine learning to neurotoxicity assessment at the International Neurotoxicology Association Conference (INA-18). This work is joint research with National Institute of Health Sciences.

Members’ Page has been updated.

Our paper has been accepted to Genetic and Evolutionary Computation Conference (GECCO) 2023. This paper introduces the margin correction method into (1+1)-CMA-ES and evaluates it on discrete and mixed-integer black-box optimization problems. This paper is joint work with M. Nomura at CyberAgent, Inc.

Our paper has been accepted to EvoApplications 2023 (Part of Evostar 2023). This work proposes a switching mechanism for target variables of the surrogate model in surrogate-assisted (1+1)-CMA-ES.

• Yutaro Yamada, Kento Uchida, Shota Saito, and Shinichi Shirakawa: Surrogate-Assisted (1+1)-CMA-ES with Switching Mechanism of Utility Functions, EvoApplications 2023 (Held as Part of EvoStar 2023), Brno, Czech Republic, April 12-14, 2023.