Kenta Oono Ph.D.

Engineering Manager at Preferred Networks, Inc

Visiting scientist at RIKEN Center for Advanced Intelligence Project (RIKEN AIP), Deep Learning Theory Team

Doctor of Philosophy in the field of Information Science and Technology

E-mail: k.oono.delta at gmail.com

HP: https://sites.google.com/view/kentaoono/

News

• 2/7/2024: We publish a paper, GenerRNA: A generative pre-trained language model for de novo RNA design [bioRxiv]

• 10/31/2023: Our paper, Universal Approximation Property of Invertible Neural Networks, has been accepted in the Journal of Machine Learning Research (JMLR)  [Journal URL]

• 6/21/2023: We publish a white paper, Virtual Human Generative Model: Masked Modeling Approach for Learning Human Characteristics [arXiv]

• 6/10/2023: Our protocol paper, Assessment of Multidimensional Health Care Parameters Among Adults in Japan for Developing a Virtual Human Generative Model: Protocol for a Cross-sectional Study, has been published at JMIR Research Protocols [Journal URL]

• 5/3/2023: Our paper, Controlling Posterior Collapse by an Inverse Lipschitz Constraint on the Decoder Network, has been accepted at ICML 2023 [arXiv]

Summary

Kenta Oono is an engineer at Preferred Networks, Inc. Japan (PFN). He engages in research on theoretical analysis of machine learning (ML) and deep learning (DL), and their application to life science technology. He received MSc. in mathematical science and a Ph.D. in information science and technology at the University of Tokyo. His research interest includes theoretical analysis of ML and DL, especially graph neural networks. He has practical experience of many data-analysis projects, especially for biological data. In addition, he has experience of development of open-source software for data mining such as Jubatus, a distributed online ML framework, and Chainer, a define-by-run DL framework.  He has performed tutorials and invited talks given lectures on theories and implementation of DL at major universities and conferences.

Professional interest

• Software development related to ML and DL

• Theoretical analysis of ML and DL

• Application of ML and DL to biology and life science

• (What is called) pure mathematics such as differential geometry, functional analysis, representation theory, and its application to ML and DL

Publication

Journal (Refereed)

1. Kenta Oono and Yuichi Yoshida, Testing properties of functions on finite groups. Random Structures & Algorithms, 49(3), 579-598, 2016, [Journal Site] [arXiv:1509:00930]

2. Naruki Yoshikawa, Kei Terayama, Teruki Honma, Kenta Oono, and Koji Tsuda, Population-based de novo molecule generation, using grammatical evolution. Chemistry Letters, 47, 1431--1434, 2018 (Editor's Choice).  [Journal Site] [arXiv:1804.02134]

3. Juntaro Matsuzaki, Ken Kato, Kenta Oono, Naoto Tsuchiya, Kazuki Sudo, Akihiko Shimomura, Kenji Tamura, Sho Shiino, Takayuki Kinoshita, Hiroyuki Daiko, Takeyuki Wada, Hitoshi Katai, Hiroki Ochiai, Yukihide Kanemitsu, Hiroyuki Takamaru, Seiichiro Abe, Yutaka Saito, Narikazu Boku, Shunsuke Kondo, Hideki Ueno, Takuji Okusaka, Kazuaki Shimada, Yuichiro Ohe, Keisuke Asakura, Yukihiro Yoshida, Shun-Ichi Watanabe, Naofumi Asano, Akira Kawai, Makoto Ohno, Yoshitaka Narita, Mitsuya Ishikawa, Tomoyasu Kato, Hiroyuki Fujimoto, Shumpei Niida, Hiromi Sakamoto, Satoko Takizawa, Takuya Akiba, Daisuke Okanohara, Kouya Shiraishi, Takashi Kohno, Fumitaka Takeshita, Hitoshi Nakagama, Nobuyuki Ota, Takahiro Ochiya, the Project Team for Development and Diagnostic Technology for Detection of miRNA in Body Fluids, Prediction of tissue-of-origin of early-stage cancers using serum miRNomes, JNCI Cancer Spectrum, 2022, pkac080, [Journal Site]

4. Masanobu Hibi, Shun Katada, Aya Kawakami, Kotatsu Bito, Mayumi Ohtsuka, Kei Sugitani, Adeline Muliandi, Nami Yamanaka, Takahiro Hasumura, Yasutoshi Ando, Takashi Fushimi, Teruhisa Fujimatsu, Tomoki Akatsu, Sawako Kawano, Ren Kimura, Shigeki Tsuchiya, Yuuki Yamamoto, Mai Haneoka, Ken Kushida, Tomoki Hideshima, Eri Shimizu, Jumpei Suzuki, Aya Kirino, Hisashi Tsujimura, Shun Nakamura, Takashi Sakamoto, Yuki Tazoe, Masayuki Yabuki, Shinobu Nagase, Tamaki Hirano, Reiko Fukuda, Yukari Yamashiro, Yoshinao Nagashima, Nobutoshi Ojima, Motoki Sudo, Naoki Oya, Yoshihiko Minegishi, Koichi Misawa, Nontawat Charoenphakdee, Zhengyan Gao, Kohei Hayashi, Kenta Oono, Yohei Sugawara, Shoichiro Yamaguchi, Takahiro Ono, Hiroshi Maruyama, Assessment of Multidimensional Health Care Parameters Among Adults in Japan for Developing a Virtual Human Generative Model: Protocol for a Cross-sectional Study, JMIR Res Protoc 2023;12:e47024, doi: 10.2196/47024, PMID: 37294611 [Journal Site]

5. Isao Ishikawa, Takeshi Teshima, Koichi Tojo, Kenta Oono, Masahiro Ikeda, and Masashi Sugiyama, Universal Approximation Property of Invertible Neural Networks, Journal of Machine Learning Research (JMLR), 24(287):1−68, 2023.

International Conference  (Refereed)

1. Kenta Oono and Taiji Suzuki, Approximation and Non-parametric Estimation of ResNet-type Convolutional Neural Networks. Proceedings of the 36th International Conference on Machine Learning (ICML 2019), 97, 4922--4931, 2019 (Acceptance ratio=773/3424=22.6%).  [arXiv:1903.10047] [PMLR] [Slide] [Poster]

2. Kenta Oono and Taiji Suzuki,  Graph Neural Networks Exponentially Lose Expressive Power for Node Classification. The eighth International Conference on Learning Representations (ICLR 2020), 2020 (Former title: On Asymptotic Behaviors of Graph CNNs from Dynamical Systems Perspective. Accepted as Spotlight, top 6.0%=(48+108)/2594, acceptance ratio=687/2594=26.5%). [arXiv:1905.10947] [OpenReview] [Slide] [Code]

3. Kenta Oono and Taiji Suzuki, Optimization and Generalization Analysis of Transduction through Gradient Boosting and Application to Multi-scale Graph Neural Networks, Advances in Neural Information Processing Systems 33 (NeurIPS 2020), 2020. (Acceptance ratio=1990/9454=21.0%) [arXiv:2006.08550] [Proceeding] [Code]

4. Takeshi Teshima, Isao Ishikawa, Koichi Tojo, Kenta Oono, Masahiro Ikeda, Masashi Sugiyama, Coupling-based Invertible Neural Networks Are Universal Diffeomorphism Approximators, Advances in Neural Information Processing Systems 33 (NeurIPS 2020), 2020. (Accepted as Oral, top 1.1%=105/9454, Acceptance ratio=1990/9454=21.0%). [arXiv:2006.11469] [Proceeding]

5. Yuri Kinoshita, Kenta Oono, Kenji Fukumizu, Yuichi Yoshida, and Shin-ichi Maeda, Controlling Posterior Collapse by an Inverse Lipschitz Constraint on the Decoder Network, Fortieth International Conference on Machine Learning (ICML 2023), 2023. (Acceptance ratio=1827/6538=27.9%) [arXiv:2304.12770]

Preprint

1. Shion Honda, Hirotaka Akita, Katsuhiko Ishiguro, Toshiki Nakanishi, and Kenta Oono, Graph Residual Flow for Molecular Graph Generation, 2019. [arXiv:1909.13521]

2. Katsuhiko Ishiguro, Kenta Oono, and Kohei Hayashi, Weisfeiler-Lehman Embedding for Molecular Graph Neural Networks, 2020. [arXiv:2006.06909]

3. Takeshi Teshima, Koichi Tojo, Masahiro Ikeda, Isao Ishikawa, and Kenta Oono, Universal Approximation Property of Neural Ordinary Differential Equations, 2020. [arXiv:2012.02414]

4. Isao Ishikawa, Takeshi Teshima, Koichi Tojo, Kenta Oono, Masahiro Ikeda, Masashi Sugiyama, Universal approximation property of invertible neural networks, 2022.[arXiv:2204.07415]

5. Kenta Oono, Nontawat Charoenphakdee, Kotatsu Bito, Zhengyan Gao, Yoshiaki Ota, Shoichiro Yamaguchi, Yohei Sugawara, Shin-ichi Maeda, Kunihiko Miyoshi, Yuki Saito, Koki Tsuda, Hiroshi Maruyama, Kohei Hayashi, Virtual Human Generative Model: Masked Modeling Approach for Learning Human Characteristics [arXiv:2306.10656]

6. Yichong Zhao,  Kenta Oono,  Hiroki Takizawa,  Masaaki Kotera, GenerRNA: A generative pre-trained language model for de novo RNA design, https://doi.org/10.1101/2024.02.01.578496  [bioRxiv]

International Workshop (Refereed)

1. Seiya Tokui,  Kenta Oono, Shohei Hido and Justin Clayton, Chainer: a Next-Generation Open Source Framework for Deep Learning, Proceedings of Workshop on Machine Learning Systems (LearningSys) in The Twenty-ninth Annual Conference on Neural Information Processing Systems (NIPS 2015), 2015 [Paper (PDF)]

2. Hai Nguyen, Shin-ichi Maeda, and Kenta Oono, Semi-supervised learning of hierarchical representations of molecules using neural message passing, Workshop on Machine Learning for Molecules and Materials in The Thirty-first Annual Conference on Neural Information Processing Systems (NIPS 2017), 2017 [Workshop Website] [arXiv:1711.10168]

3. Rina Onda, Zhengyan Gao, Masaaki Kotera, Kenta Oono, Fast Estimation Method for the Stability of Ensemble Feature Selectors, SubSetML: Subset Selection in Machine Learning: From Theory to Practice in The 38th International Conference on Machine Learning (ICML 2021), 2021 [Workshop Website] [arXiv:2108.01485]

4. Soma Onishi, Kenta Oono, Kohei Hayashi, TabRet: Pre-training Transformer-based Tabular Models for Unseen Columns, Workshop on Mathematical and Empirical Understanding of Foundation Models (ME-FoMo) at The Eleventh International Conference on Learning Representations (ICLR 2023), 2023 [OpenReview] [arXiv]

Journal (Non-refereed)

1. Kenta Oono, Over-Smoothing of Graph Neural Networks, The Japanese Society for Artificial Intelligence, Artificial Intelligence Vol.38 No.2, 149--157, 2023 (Japanese) [Journal URL]

Talks

International Conference (Tutorials)

1. Seiya Tokui, Kenta Oono, and Atsunori Kanemura,  Deep Learning Implementations and Frameworks, the Thirty-first AAAI Conference on Artificial Intelligence (AAAI), 2017, [Tutorial Website]

2. Seiya Tokui, Kenta Oono, Atsunori Kanemura, and Toshihiro Kamishima, Deep Learning Implementations and Frameworks, the 20th Pacific Asia Conference on Knowledge Discovery and Data Mining (PAKDD), 2016 [Tutorial Website]

International Conference/Workshop (Invited talk)

1. Kenta Oono, Recent Development of Deep Learning Technology and its Application to Quantitative Structure-Activity Relationship, Joint conference of International Conference on Genome Informatics (GIW) and International Conference on Bioinformatics (InCoB), 2015.

2. Kenta Oono, Explaining Oversmoothing of Non-linear Graph Neural Networks. The 11th Asian Conference on Machine Learning (ACML), Workshop on Statistics & Machine Learning Researchers in Japan (StatsML Japan), 2019 [Workshop Website].

International conference (Contributed talk, Refereed)

1. Kenta Oono and Taiji Suzuki, Approximation and Non-parametric Estimation of ResNet-type Convolutional Neural Networks. The 36th International Conference on Machine Learning (ICML 2019). 2019 Jun.

2. Kenta Oono and Taiji Suzuki, Toward Understanding Expressive Power of Graph Convolutional Neural Networks, Data Science, Statistics & Visualisation (DSSV), 2019. [URL]

3. Kenta Oono and Taiji Suzuki, Graph Neural Networks Exponentially Lose Expressive Power for Node Classification. The eighth International Conference on Learning Representations (ICLR 2020), 2020 Apr.

4. Kenta Oono and Taiji Suzuki, Optimization and Generalization Analysis of Transduction through Gradient Boosting and Application to Multi-scale Graph Neural Networks, Thirty-fourth Conference on Neural Information Processing Systems (NeurIPS2020), 2020 Dec.

International conference (Contributed talk, Non-refereed)

1. Kenta Oono, Speed up deep learning R&D with Chainer, AAAI 2017 Spring Symposia, Wellbeing AI: From Machine Learning to Subjectivity Oriented Computing, 2017

Lectures

• UC San Francisco, BMI219, Deep Learning (2017) [URL], [code1], [code2]

• The University of Tokyo, Frontier Artificial Intelligence II (guest speaker 2016, 2017) , [URL]

Award

• Dean's award from the University of Tokyo for research in the doctoral course [News].

• Presentation award in IBISML 2020 [Workshop Website (Japanese)].

• Top 33% reviewer at ICML 2020 [Certificate]

• Outstanding reviewer at ICLR2021 (473 / 4072 reviewers = 11.6%)

• Outstanding  reviewer at NeurIPS 2021 (Top 8%) [Official website]

• Top reviewer at AISTATS 2022 (Top 10%) [Official website]

• Highlighted Reviewer at ICLR 2022 [Official website]

Community Activities

• Reviewer

  • ICLR (2020, 2021, 2022, 2023)

  • ICML (2020, 2021, 2022, 2023)

  • NeurIPS (2020, 2021, 2022)

  • AISTATS (2021, 2022, 2023)

  • AAAI (2021, 2022)

  • Neural Networks (2019, 2020, 2021, 2022, 2023)

  • IEEE Transactions on Pattern Analysis and Machine Intelligence (2019, 2020, 2021)

  • IEEE Transactions on Neural Networks and Learning Systems (2023)

  • Neurocomputing (2021)

  • Journal of Approximation Theory (2021)

  • Annals of the Institute of Statistical Mathematics (2021)

  • Information Fusion (2021, 2022)

  • Journal of Statistical Planning and Inference (2022)

  • Transactions on Machine Learning Research (2022, 2023)

• Japan MENSA member

• Japanese Association for Medical Artificial Intelligence,  Councilor (2018--) [HP (Japanese)]

• Mathcafe Japan (non-profit organization), Director (2022--) [HP (Japanese)]

Employment history

Oct. 2014 - Current : Preferred Networks, Inc. Japan

• Bio project team. Lead the team and conducted data-analysis and prototyping projects.

• Development of a deep learning framework, Chainer. Member of core development team which is in charge of making development policies, implement of new features, review and so on. 

• Recruiting team. Lead two-month long internship project (selection of candidates, preparation, and execution)

Apr. 2011 - Oct. 2014 : Preferred Infrastructure, Inc. Japan

• ML related research and development

• Pre-sales and support engineer of proprietary software

• Development of Jubatus, a distributed online machine learning framework

• Bio project leader, whose goal is research and development of ML for biohealthcare and its commercialization.

Aug. 2011- Sep. 2011: Preferred Infrastructure, Inc. Japan (Intern)

• Developed algorithms for burst detection from microblog (e.g. Twitter) streams.

Educational history

• Apr. 2018 - Mar. 2021: The University of Tokyo, Graduate School of Information Science and Technology [HP], Doctor course

  • Thesis title: Analysis of Deep Learning from the Viewpoint of Model Structures [UTokyo Repository]

  • Supervisor: Dr. Taiji Suzuki [HP]

• Apr. 2009 - Mar. 2011: The University of Tokyo, Graduate School of Mathematical Science [HP], Master course

  • Thesis title: Conformally Invariant Operators on Symmetric Tensor Fields

  • Supervisor: Dr. Kengo Hirachi [HP]

• Apr. 2007 - Mar. 2009: The University of Tokyo, School of Science, Department of Mathematics [HP]

• Apr. 2005 - Mar. 2007: The University of Tokyo, College of Arts and Sciences, Natural Sciences I [HP]

• Apr. 2002 - Mar. 2005: Kaisei high school [HP (Japanese) ]

Patents

• US patent : US2017/0161635A1, Generative Machine Learning Systems for Drug Design

• Other 5 patents are pending.

Qualification

• TOEIC : 970/990 (Dec. 2018)

• TOEFL iBT : 94/120 (Oct. 2017)

• Abacus: Quasi first dan

Skill

• Programming language: Python, C++, Ruby (ordered by proficiency)

• Japanese (native), English (business)

Social activities

• Twitter (Japanese) : @delta2323_

• Facebook (Japanese): kenta.oono.1

• Slideshare (Japanese) : Kenta Oono

• Github : delta2323

Hobby

• Workout, Running (half marathon best time: 1h 39min.)