Hello World!

Ryuichiro Hataya’s webpage.

About

profile.png

I am Ryuichiro Hataya, PhD (Information Science and Technology).

  • Postdoctoral researcher @ RIKEN Advanced Data Science Project
  • Postdoctoral researcher @ RIKEN Center for Advanced Intelligence Project
  • Visiting researcher @ International Center for Elementary Particle Physics, UTokyo
  • Gradient-based hyperparameter optimization
  • Operator-theoretic machine learning
  • Applications of machine learning to other research fields (medical imaging, palaeontology, chemistry, physics, etc.)
  • I will visit Vietnam Institute for Advanced Study in Mathematics in April.
  • I will visit EPFL CIS (Switzerland) and Fraunhofer IIS (Germany) from 8th to 15th March 2023.
  • Our paper “Nyström Method for Accurate and Scalable Implicit Differentiation” has been accepted at AISTATS 2023.
  • I joined RIKEN ADSP and RIKEN AIP as a postdoctral researcher
  • I recieved a doctal degree as a representative student of the graduate school of Information Science and Techonology, UTokyo.
  • I defended my PhD thesis.
  • I visited IIT (Genova, Italy) from July 8th.
  • Our paper “DJMix: Unsupervised Task-agnostic Image Augmentation for Improving Robustness of Convolutional Neural Networks” is accepted to IJCNN 2022.
older news
  • I visited IIT (Genova, Italy) from October 1st to December 17th.
  • Our paper “Meta Approach to Data Augmentation Optimization” is accepted to WACV 2022.
  • My research proposal has been accepted in JST’s ACT-X.
  • Call for NeurIPS meetups is now out!
  • I will present about Faster AutoAugment and its applications at AIP Open seminar.
  • Our paper “Graph Energy-based Model for Molecular Graph Generation” is accepted at EBM workshop 2021 as a contributed talk.
  • I will serve as a meetup chair for NeurIPS 2021.
  • My research proposal has been accepted in JSPS’s travel grant.
  • My research proposals have been accepted by Microsoft Research Asia, and RIISE at UTokyo.
  • We organized a NeurIPS meetup and Women in ML in Japan: https://neuripsmeetupjapan.github.io.
  • Our paper “Decomposing Normal and Abnormal Features of Medical Images for Content-based Image Retrieval” is accepted at ML4H 2020.

Projects


Latest Post

All Posts

Publications

bird.png
  • Ryuichiro Hataya, and Makoto Yamada, “Nyström Method for Accurate and Scalable Implicit Differentiation,” International Conference on Artificial Intelligence and Statistics, Spain, 2023.
  • Ryuichiro Hataya, and Hideki Nakayama, “DJMix: Unsupervised Task-agnostic Image Augmentation for Improving Robustness of Convolutional Neural Networks,” International Joint Conference on Neural Networks, 2022.
  • Ryuichiro Hataya, Jan Zdenek, Kazuki Yoshizoe, and Hideki Nakayama, “Meta Approach to Data Augmentation Optimization.” Winter Conference on Applications of Computer Vision, 2022.
  • Taiga Kashima, Ryuichiro Hataya, and Hideki Nakayama, “Visualizing Association in Exemplar-based Classification.” International Conference on Acoustics, Speech, and Signal Processing, 2021.
  • Ryuichiro Hataya, Jan Zdenek, Kazuki Yoshizoe, and Hideki Nakayama, “Faster AutoAugment: Learning Augmentation Strategies using Backpropagation.” European Conference on Computer Vision, 2020.
  • Ryuichiro Hataya, and Hideki Nakayama, “LOL: Learning To Optimize Loss Switching Under Label Noise.” International Conference on Image Processing, 2019.
Preprints and others
  • Ryuichiro Hataya, and Yuka Hashimoto, “Noncommutative $C^*$-algebra Net: Learning Neural Networks with Powerful Product Structure in $C^*$-algebra,” 2023. arXiv
  • Ryuichiro Hataya, Bao Han, and Hiromi Arai, “Will Large-scale Generative Models Corrupt Future Datasets?” 2022. arXiv
  • Ryuichiro Hataya, Hideki Nakayama, and Kazuki Yoshizoe, “Graph Energy-based Model for Substructure Preserving Molecular Design,” 2021. arxiv
  • Ryuichiro Hataya, Hideki Nakayama, and Kazuki Yoshizoe, “Graph Energy-based Model for Molecular Graph Generation.” EBM Workshop at ICLR 2021, 2021. (Peer Reviewed, Contributed Talk)
  • Kazuma Kobayashi, Ryuichiro Hataya, Yusuke Kurose, Tatsuya Harada, and Ryuji Hamamoto, “Decomposing Normal and Abnormal Features of Medical Images for Content-based Image Retrieval.” Machine Learning for Health Workshop at NeurIPS 2020. (Peer Reviewed, Extended Abstract)
  • Ryuichiro Hataya, Kumiko Matsui, and Tomoki Karasawa, “Learning to Identify Large Fossils using Deep Convolutional Neural Networks”, Geological Society of America Abstracts with Programs. Vol 52, No. 6, 2020.
  • Ryuichiro Hataya, and Hideki Nakayama, “Unifying semi-supervised and robust leaning by mixup.” Workshop on Learning from Limited Labeled Data at ICLR 2019, 2019. (Peer Reviewed, Spotlight)

Other Research Activities

prague.png
  • “Bayesian Model Selection for Deep Learning” (in Japanese), Zapping Seminar, 2022.
  • “Data Augmentation for Deep Learning” (in Japanese), Ehime University DS Research Seminar, 2021.
  • “Data Augmentation for Deep Learning” (in Japanese), Symposium on Sensing via Image Information, 2021.
  • “Data Augmentation for Deep Learning” (in Japanese), StatsML Symposium, 2020.
  • “Gradient-based Hyperparameter Optimization” (in Japanese), Zapping Seminar, 2020.
  • Research Project of Interactive Data Generation, Japan Science and Technology Agency, ACT-X, ¥4.5M, 2021-2024.
  • Research Project of Differentiable Data Augmentation for Image Recognition, Overseas Challenge Program for Young Researchers by JSPS, ¥1.4M, 2021.
  • Research Project of Interactive Image Generation, Microsoft Research Asia Collaborative Research Program (D-CORE 2021) by MSRA, ¥1.0M, 2021.
  • Research Project of Inclusive Image Recognition, Sprouting Research RA’s in Value Exchange Engineering by RIISE@UTokyo, ¥2.0M, 2020-2022.
  • Microsoft Research Asia D-Core Award, 2020.
  • Best Student Paper Award, The 23rd Meeting on Image Recognition and Understanding, 2020.
  • Meetup Chair of NeurIPS, 2021.
  • Organizer of NeurIPS meetup Japan & Women in ML, 2020.
  • Volunteer for ICML, and ICLR, 2020.
  • Reviewer for NeurIPS, ICCV, ICLR, CVPR, ICML 2019~.