Updated on 2024/04/04

写真a

 
TAKENO Shion
 
Organization
Graduate School of Engineering Mechanical Systems Engineering 2 Assistant Professor
Undergraduate School
School of Engineering Mechanical and Aerospace Engineering
Title
Assistant Professor

Research Interests 4

  1. Materials Informatics

  2. Bayesian optimization

  3. Bioinformatics

  4. Gaussian process bandit

Research History 7

  1. Nagoya University   Graduate School of Engineering   Assistant Professor

    2024.4

      More details

    Country:Japan

  2. Japan Science and Technology Agency   ACT-X   Researcher

    2023.10

      More details

    Country:Japan

  3. Nagoya University   Graduate School of Engineering   Invited Researcher

    2023.4 - 2024.3

      More details

    Country:Japan

  4. RIKEN   Data-Driven Biomedical Science Team   Postdoctoral Researcher

    2023.4 - 2024.3

      More details

    Country:Japan

  5. CyberAgents   AI Lab   Collaborative Researcher

    2022.6 - 2023.3

      More details

    Country:Japan

  6. Japan Society for the Promotion of Science   Research Fellowship for Young Scientists (DC2)

    2021.4 - 2023.3

      More details

    Country:Japan

  7. RIKEN   Data-Driven Biomedical Science Team   Junior Research Associate

    2020.4 - 2021.3

      More details

    Country:Japan

▼display all

Education 2

  1. Nagoya Institute of Technology   Graduate School of Engineering Doctor's Course   Computer Science and Engineering

    2020.4 - 2023.3

      More details

    Country: Japan

  2. Nagoya Institute of Technology   Graduate School of Engineering Master's course   Computer Science and Engineering

    2018.4 - 2020.3

      More details

    Country: Japan

Awards 5

  1. IBIS Workshop 優秀プレゼンテーション賞 ファイナリスト

    2023.11  

  2. IBISML 研究会賞 ファイナリスト

    2023.10  

  3. Nagoya Institute of Technology Student Research Encouragement Award by President

    2023.3   Nagoya Institute of Technology  

  4. IEICE TC-IBISML Research Award Finalist (Co-author)

    2022.10  

  5. Nagoya Institute of Technology Student Research Encouragement Award by Vice-president

    2021.3   Nagoya Institute of Technology  

 

Papers 15

  1. Randomized Gaussian Process Upper Confidence Bound with Tighter Bayesian Regret Bounds Reviewed

    Shion Takeno, Yu Inatsu, Masayuki Karasuyama

    Proceedings of the 40th International Conference on Machine Learning (ICML)   Vol. 202   page: 33490 - 33515   2023

     More details

    Authorship:Lead author   Language:English   Publishing type:Research paper (international conference proceedings)  

  2. Towards Practical Preferential Bayesian Optimization with Skew Gaussian Processes Reviewed

    Shion Takeno, Masahiro Nomura, Masayuki Karasuyama

    Proceedings of the 40th International Conference on Machine Learning (ICML)   Vol. 202   page: 33516 - 33533   2023

     More details

    Authorship:Lead author, Corresponding author   Language:English   Publishing type:Research paper (international conference proceedings)  

  3. Sequential and Parallel Constrained Max-value Entropy Search via Information Lower Bound. Reviewed

    Shion Takeno, Tomoyuki Tamura, Kazuki Shitara, Masayuki Karasuyama

    Proceedings of the 39th International Conference on Machine Learning (ICML)     page: 20960 - 20986   2022

     More details

    Authorship:Lead author   Language:English   Publishing type:Research paper (international conference proceedings)  

    Other Link: https://dblp.uni-trier.de/rec/conf/icml/2022

  4. Multi-fidelity Bayesian Optimization with Max-value Entropy Search and its Parallelization. Reviewed

    Shion Takeno, Hitoshi Fukuoka, Yuhki Tsukada, Toshiyuki Koyama, Motoki Shiga, Ichiro Takeuchi, Masayuki Karasuyama

    Proceedings of the 37th International Conference on Machine Learning (ICML)     page: 9334 - 9345   2020

     More details

    Authorship:Lead author   Language:English   Publishing type:Research paper (international conference proceedings)   Publisher:PMLR  

    Other Link: https://dblp.uni-trier.de/rec/conf/icml/2020

  5. Bounding Box-based Multi-objective Bayesian Optimization of Risk Measures under Input Uncertainty Reviewed

    Yu Inatsu, Shion Takeno, Hiroyuki Hanada, Kazuki Iwata, Ichiro Takeuchi

    Proceedings of the 27th International Conference on Artificial Intelligence and Statistics (to appear)     2024

     More details

    Language:English   Publishing type:Research paper (international conference proceedings)  

  6. Risk Seeking Bayesian Optimization under Uncertainty for Obtaining Extremum Reviewed

    Shogo Iwazaki, Tomohiko Tanabe, Mitsuru Irie, Shion Takeno, Yu Inatsu

    Proceedings of the 27th International Conference on Artificial Intelligence and Statistics (to appear)     2024

     More details

    Language:English   Publishing type:Research paper (international conference proceedings)  

  7. Multi-objective Bayesian Optimization with Active Preference Learning Reviewed

    Ryota Ozaki, Kazuki Ishikawa, Youhei Kanzaki, Shinya Suzuki, Shion Takeno, Ichiro Takeuchi, Masayuki Karasuyama

    Proceedings of the 38th AAAI Conference on Artificial Intelligence (to appear)     2024

     More details

    Language:English  

  8. Failure-Aware Gaussian Process Optimization with Regret Bounds Reviewed

    Shogo Iwazaki, Shion Takeno, Tomohiko Tanabe, Mitsuru Irie

    Advances in Neural Information Processing Systems 36 (NeurIPS 2023)     page: 24388 - 24400   2023

     More details

    Language:English   Publishing type:Research paper (international conference proceedings)  

  9. A Generalized Framework of Multifidelity Max-Value Entropy Search Through Joint Entropy. Reviewed

    Shion Takeno, Hitoshi Fukuoka, Yuhki Tsukada, Toshiyuki Koyama, Motoki Shiga, Ichiro Takeuchi, Masayuki Karasuyama

    Neural Computation   Vol. 34 ( 10 ) page: 2145 - 2203   2022

     More details

    Authorship:Lead author   Language:English   Publishing type:Research paper (scientific journal)  

    DOI: 10.1162/neco_a_01530

  10. Bayesian Optimization for Cascade-Type Multistage Processes. Reviewed

    Shunya Kusakawa, Shion Takeno, Yu Inatsu, Kentaro Kutsukake, Shogo Iwazaki, Takashi Nakano, Toru Ujihara, Masayuki Karasuyama, Ichiro Takeuchi

    Neural Computation   Vol. 34 ( 12 ) page: 2408 - 2431   2022

     More details

    Language:English   Publishing type:Research paper (scientific journal)  

    DOI: 10.1162/neco_a_01550

  11. Preferential Bayesian Optimization with Hallucination Believer Reviewed

    Shion Takeno, Masahiro Nomura, Masayuki Karasuyama

    NeurIPS Workshop on Gaussian Processes, Spatiotemporal Modeling, and Decision-making Systems     2022

     More details

    Authorship:Lead author   Language:English  

  12. Bayesian Optimization for Distributionally Robust Chance-constrained Problem. Reviewed

    Yu Inatsu, Shion Takeno, Masayuki Karasuyama, Ichiro Takeuchi

    Proceedings of the 39th International Conference on Machine Learning (ICML)     page: 9602 - 9621   2022

     More details

    Language:English   Publishing type:Research paper (international conference proceedings)  

    Other Link: https://dblp.uni-trier.de/rec/conf/icml/2022

  13. Cost-effective search for lower-error region in material parameter space using multifidelity Gaussian process modeling Reviewed

    Shion Takeno, Yuhki Tsukada, Hitoshi Fukuoka, Toshiyuki Koyama, Motoki Shiga, Masayuki Karasuyama

    Physical Review Materials   Vol. 4 ( 8 )   2020.8

     More details

    Authorship:Lead author   Language:English   Publishing type:Research paper (scientific journal)  

    Information regarding precipitate shapes is critical for estimating material parameters. Hence, we considered estimating a region of material parameter space in which a computational model produces precipitates having shapes similar to those observed in the experimental images. This region, called the lower-error region (LER), reflects intrinsic information of the material contained in the precipitate shapes. However, the computational cost of LER estimation can be high because the accurate computation of the model is required many times to better explore parameters. To overcome this difficulty, we used a Gaussian-process-based multifidelity modeling, in which training data can be sampled from multiple computations with different accuracy levels (fidelity). Lower-fidelity samples may have lower accuracy, but the computational cost is lower than that for higher-fidelity samples. Our proposed sampling procedure iteratively determines the most cost-effective pair of a point and a fidelity level for enhancing the accuracy of LER estimation. We demonstrated the efficiency of our method through estimation of the interface energy and lattice mismatch between MgZn2 and α-Mg phases in an Mg-based alloy. The results showed that the sampling cost required to obtain accurate LER estimation could be drastically reduced.

    DOI: 10.1103/PhysRevMaterials.4.083802

    Scopus

  14. Multi-objective Bayesian Optimization using Pareto-frontier Entropy. Reviewed

    Shinya Suzuki, Shion Takeno, Tomoyuki Tamura, Kazuki Shitara, Masayuki Karasuyama

    Proceedings of the 37th International Conference on Machine Learning(ICML)     page: 9279 - 9288   2020

     More details

    Language:English   Publishing type:Research paper (international conference proceedings)   Publisher:PMLR  

    Other Link: https://dblp.uni-trier.de/rec/conf/icml/2020

  15. Estimation of material parameters based on precipitate shape: efficient identification of low-error region with Gaussian process modeling Reviewed

    Yuhki Tsukada, Shion Takeno, Masayuki Karasuyama, Hitoshi Fukuoka, Motoki Shiga, Toshiyuki Koyama

    Scientific Reports   Vol. 9 ( 1 )   2019.12

     More details

    Language:English   Publishing type:Research paper (scientific journal)  

    In this study, an efficient method for estimating material parameters based on the experimental data of precipitate shape is proposed. First, a computational model that predicts the energetically favorable shape of precipitate when a d-dimensional material parameter (x) is given is developed. Second, the discrepancy (y) between the precipitate shape obtained through the experiment and that predicted using the computational model is calculated. Third, the Gaussian process (GP) is used to model the relation between x and y. Finally, for identifying the “low-error region (LER)” in the material parameter space where y is less than a threshold, we introduce an adaptive sampling strategy, wherein the estimated GP model suggests the subsequent candidate x to be sampled/calculated. To evaluate the effectiveness of the proposed method, we apply it to the estimation of interface energy and lattice mismatch between MgZn2 (β1') and α-Mg phases in an Mg-based alloy. The result shows that the number of computational calculations of the precipitate shape required for the LER estimation is significantly decreased by using the proposed method.

    DOI: 10.1038/s41598-019-52138-0

    Scopus

    PubMed

▼display all

Books 2

  1. マテリアル・機械学習・ロボット(現代化学増刊48)

    竹野思温, 烏山昌幸, 竹内一郎( Role: Contributor ,  第Ⅲ部 材料科学のためのベイズ最適化 使いこなし, 9 精度・正確度と観測コストを考慮したマルチフィデリティ最適化)

    東京化学同人  2024.3  ( ISBN:9784807913480

  2. マテリアルズインフォマティクスのためのデータ作成とその解析、応用事例

    執筆者:58名(第4章 第5節 精度と観測コストのトレードオフを考慮したベイズ的探索法: 材料パラメータ推定での適用事例)

    技術情報協会  2023.10  ( ISBN:4861048540

     More details

    Total pages:500   Language:Japanese

    CiNii Books

    ASIN

MISC 1

  1. ベイズ最適化の基礎と材料探索への応用

    Materials stage     2023.6

Presentations 20

  1. 選好考慮型多目的ベイズ最適化によるユーザの好みを反映したハイパーパラメータ最適化

    尾崎令拓, 石川和樹, 神崎陽平, 竹野思温, 竹内一郎, 烏山昌幸

    情報論的学習理論と機械学習研究会(IBISML2022)  2022.12 

     More details

    Presentation type:Oral presentation (general)  

  2. 複数目的のレベル集合推定のための適応的意思決定アルゴリズムの提案

    岩崎 省吾, 竹野思温, 稲津佑, 松井孝太

    第37回人工知能学会全国大会  2023.6 

     More details

    Presentation type:Oral presentation (general)  

  3. 最適値の情報量に基づいたマルチフィデリティベイズ最適化

    竹野思温, 烏山昌幸

    第37回人工知能学会全国大会  2023.6 

     More details

    Presentation type:Oral presentation (general)  

  4. 情報量の下界に基づく逐次的及び並列的な制約付きベイズ最適化

    竹野思温, 田村友幸, 設楽一希, 烏山昌幸

    情報論的学習理論と機械学習研究会(IBISML2022)  2022.1 

     More details

    Presentation type:Oral presentation (general)  

  5. 情報量の下界に基づく制約付きベイズ最適化

    竹野思温, 田村友幸, 設楽一希, 烏山昌幸

    第24回情報論的学習理論ワークショップ (IBIS2021)  2021.11 

     More details

    Presentation type:Oral presentation (general)  

  6. 情報量に基づく複数タスクの同時ベイズ最適化

    山田倫太郎, 竹野思温, 烏山昌幸

    情報論的学習理論と機械学習研究会(IBISML2022)  2022.1 

     More details

    Presentation type:Oral presentation (general)  

  7. 専門家の知識を入れたものづくりのためのデータ同化(i)-手法の提案-

    沓掛健太朗, 竹野思温, 太田壮音, 烏山昌幸, 竹内一郎, 宇治原徹

    第83回応用物理学会秋季学術講演会  2022.9 

  8. 専門家の知識を入れたものづくりのためのデータ同化(ii)-SiC溶液成長シミュレーションへの適用-

    太田壮音, 沓掛健太朗, 竹野思温, 烏山昌幸, 竹内一郎, 原田俊太, 田川美穂, 宇治原徹

    第83回応用物理学会秋季学術講演会  2022.9 

  9. 多段階プロセスに対するベイズ最適化の中断可能設定への拡張

    草川隼也, 竹野思温, 沓掛健太朗, 竹内一郎

    第18回情報学ワークショップ (WINF2020)  2020.11 

     More details

    Presentation type:Oral presentation (general)  

  10. 分布的ロバストな機会制約付き最適化問題に対する能動学習

    稲津佑, 竹野思温, 烏山昌幸, 竹内一郎

    情報論的学習理論と機械学習研究会(IBISML2021)  2021.6 

     More details

    Presentation type:Oral presentation (general)  

  11. 出力空間の情報量を用いたベイズ最適化とその発展 Invited

    竹野思温

    第5回 統計・機械学習若手シンポジウム (StatsML Symposium'20)  2020.12 

     More details

    Presentation type:Public lecture, seminar, tutorial, course, or other speech  

  12. 出力空間の情報量に基づくベイズ最適化の発展

    竹野思温

    理研AIP データ駆動型生物医科学チーム オンラインセミナー  2022.9 

     More details

    Presentation type:Public lecture, seminar, tutorial, course, or other speech  

  13. 入力不確実性が存在する下での信用領域を用いたリスク尺度に対する多目的ベイズ最適化手法

    稲津佑, 花田博幸, 岩田和樹, 竹野思温, 竹内一郎

    情報論的学習理論と機械学習研究会(IBISML2023)  2023.10.30 

  14. 入力データ分布に対する不確実性の下での分布ロバストな能動学習

    大藏芳斗, 稲津佑, 竹野思温, 花田博幸, 青山竜也, 田中智成, 赤羽智志, 小嶋信矢, 李翰柱, 竹内一郎

    情報論的学習理論と機械学習研究会(IBISML2023)  2023.10.30 

  15. 事後分布からのサンプルに基づくベイズ最適化

    竹野思温, 稲津佑, 烏山昌幸, 竹内一郎

    情報論的学習理論と機械学習研究会(IBISML2023)  2023.10.30 

     More details

    Presentation type:Poster presentation  

  16. 乱択GP-UCBアルゴリズムのリグレット解析

    竹野思温, 稲津佑, 烏山昌幸

    情報論的学習理論と機械学習研究会(IBISML2022)  2022.12 

     More details

    Presentation type:Oral presentation (general)  

  17. ガウス過程に対する乱択UCBアルゴリズム

    Japanese Joint Statistical Meeting.  2023.9.5 

     More details

    Language:Japanese   Presentation type:Oral presentation (general)  

  18. カスケードタイプの多ステージプロセスに対するベイズ最適化

    草川隼也, 竹野思温, 稲津佑, 沓掛健太朗, 岩崎省吾, 中野高志, 烏山昌幸, 宇治原徹, 竹内一郎

    第24回情報論的学習理論ワークショップ (IBIS2021),  2021.11 

     More details

    Presentation type:Oral presentation (general)  

  19. Pareto-frontier Entropyに基づく多目的ベイズ最適化

    鈴木進也, 竹野思温, 田村友幸, 設楽一希, 烏山昌幸

    第22回情報論的学習理論ワークショップ (IBIS2019)  2019.11 

     More details

    Presentation type:Poster presentation  

  20. Max-value Entropy Searchに基づくMulti-fidelityベイズ最適化

    竹野思温, 福岡準史, 塚田祐貴, 小山敏幸, 志賀元紀, 竹内一郎, 烏山昌幸

    第22回情報論的学習理論ワークショップ (IBIS2019)  2019.11 

     More details

    Presentation type:Poster presentation  

▼display all

KAKENHI (Grants-in-Aid for Scientific Research) 3

  1. Developing practical human-in-the-loop preferential optimization method and its theoretical guarantee

    Grant number:JPMJAX23CD  2023.10 - 2026.3

    Japan Science and Technology Agency 

      More details

    Authorship:Principal investigator 

  2. 乱択ベイズ最適化法の開発およびその理論保証と材料分野への応用

    Grant number:23K19967  2023.8 - 2025.3

    日本学術振興会  科学研究費助成事業  研究活動スタート支援

    竹野 思温

      More details

    Authorship:Principal investigator 

    Grant amount:\2860000 ( Direct Cost: \2200000 、 Indirect Cost:\660000 )

  3. 出力空間情報量に基づくマルチフィデリティベイズ最適化とその材料分野への応用

    Grant number:21J14673  2021.4 - 2023.3

    日本学術振興会  科学研究費助成事業 特別研究員奨励費  特別研究員奨励費

    竹野 思温

      More details

    Authorship:Principal investigator 

    Grant amount:\1500000 ( Direct Cost: \1500000 )

    本年度は, 出力空間情報量に基づくマルチフィデリティベイズ最適化について拡張を行った. まず, 提案法に必要となる数値積分などの近似方法についてより効率的な計算法を確立した. これにより, 近似精度を高めつつさらに高速な情報量の評価が可能となった. さらに, より複雑な, 複数のデータの観測を同時に行える場合を考えた. 例えば, 計算材料科学などの分野では, 長い時間をかけてシミュレーションを行うことで物性値などを計算することがある. このような場合には, その途中で近似的な物性値も同時に観測することができる. 本拡張ではこのような複数のデータが同時に与えられることを考慮した指標を設計した. また, この指標の非常に効率的な導出法を示し, より効率的な最適化が行えることを示した. 本研究成果は国際雑誌へと投稿を行い, Major Revision の判定を受け修正中である.
    また, 実践上重要な制約付き最適化問題にも取り組み, 出力空間情報量に基づく制約付きベイズ最適化の研究も行った. 本研究では,制約付き最適化問題に対して既存法を単純に拡張したアプローチを適用すると理論・実践的問題が生じることを示した. この問題点に対し, 情報量の下界に基づくより頑健な近似方法を提案し, また並列観測が行える場合への拡張も行った. さらに, 得られた推定量に関する理論的な検討も行った. 既存の出力空間情報量に基づく単純な拡張を含むいくつかの手法と比較し, 高い性能を持つことを数値実験により示した. この研究成果は国際学会へと投稿を行っている.

Industrial property rights 1

  1. Estimation device, estimation method, and program

     More details

    Application no:特願2022-134361  Date applied:2022.8

 

Academic Activities 2

  1. Reviewer at The 26th International Conference on Artificial Intelligence and Statistics

    Role(s):Peer review

     More details

    Type:Academic society, research group, etc. 

  2. Reviewer at Advances in Neural Information Processing Systems 36 (NeurIPS 2023, Top Reviewer)

    Role(s):Peer review

     More details

    Type:Academic society, research group, etc.