about
Bio
PhD (Statistical Sciences) from the Institute of Statistical Mathematics with a doctoral thesis entitled “Identification of Importance-Weighting and Geodesics on Statistical Manifolds”.
- Research Fellow in statistics at The University of Melbourne
- Email: m.kimura[at]unimelb.edu.au
Selected publications
Information geometry and machine learning
- Masanari Kimura and Hideitsu Hino. “Information Geometrically Generalized Covariate Shift Adaptation.” Neural Computation 34.9 (2022): 1944-1977. paper
- Masanari Kimura and Hideitsu Hino. “α-geodesical skew divergence.” Entropy 23.5 (2021): 528. paper
- Masanari Kimura. “Generalized t-SNE Through the Lens of Information Geometry.” IEEE Access 9 (2021): 129619-129625. paper
Theoretical analysis of data augmentation
- Masanari Kimura. “Generalization Bounds for Set-to-Set Matching with Negative Sampling.” International Conference on Neural Information Processing (ICONIP 2022). paper
- Masanari Kimura. “Understanding test-time augmentation.” International Conference on Neural Information Processing (ICONIP 2021). paper
- Masanari Kimura. “Why mixup improves the model performance.” International Conference on Artificial Neural Networks (ICANN 2021). paper
Distribution shift and active learning
- Masanari Kimura and Hideitsu Hino. “A Short Survey on Importance Weighting for Machine Learning”. Transactions on Machine Learning Research. (TMLR 2024). paper
- Masanari Kimura, Takuma Nakamura and Yuki Saito. “SHIFT15M: Fashion-specific dataset for set-to-set matching with several distribution shifts.” IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshop (CVPRW 2023). paper
- Masanari Kimura, Kei Wakabayashi, and Atsuyuki Morishima. “Batch prioritization of data labeling tasks for training classifiers.” AAAI Conference on Human Computation and Crowdsourcing (AAAI HCOMP 2020). paper