Organization
Institute of Integrated Research Supercomputing Research Center Professor
Title
Professor
External link
YOKOTA RIO
Degree
-
Ph.D. (Engineering) ( Keio University )
Research Interests
-
Numerical Analysis
-
Deep Learning
-
High Performance Computing
-
GPU
Research Areas
-
Manufacturing Technology (Mechanical Engineering, Electrical and Electronic Engineering, Chemical Engineering) / Fluid engineering
-
Informatics / Intelligent informatics
-
Informatics / Mathematical informatics
-
Informatics / Computational science
-
Informatics / High performance computing
Education
-
Keio University Graduate School of Science and Technology School of Science for Open and Environmental Systems (PhD)
2005.4 - 2009.3
-
Keio University Graduate School of Science and Technology School of Science for Open and Environmental Systems (Masters)
2003.4 - 2005.3
Country： Japan
-
Keio University Faculty of Science and Technology Department of Mechanical Engineering
1997.4 - 2003.3
Country： Japan
Research History
-
RIKEN Center for Computational Science, AI for Science Platform Division, AI for Science Foundation Model Research Team Team Principal
2025.4
-
Institute of Science Tokyo Institute of Integrated Research, Supercomputing Research Center Professor
2024.10
-
National Institute of Informatics Research and Development Center for Large Language Models Scientific Director
2024.4
-
Tokyo Institute of Technology Global Scientific Information and Computing Center Professor
2023.1 - 2024.9
-
Tokyo Institute of Technology Global Scientific Information and Computing Center Associate Professor
2015.4 - 2022.12
-
King Abdullah University of Science and Technology Applied Mathematics and Computer Science Research Scientist
2011.9 - 2015.3
-
Boston University Department of Mechanical Engineering Post-doctoral Research Associate
2010.9 - 2011.8
-
University of Bristol Department of Mathematics Post-doctoral Research Associate
2009.2 - 2010.8
Professional Memberships
-
THE JAPAN SOCIETY FOR INDUSTRIAL AND APPLIED MATHEMATICS
-
THE JAPAN SOCIETY FOR COMPUTATIONAL ENGINEERING AND SCIENCE
-
INFORMATION PROCESSING SOCIETY OF JAPAN
-
Ameriacn Society of Mechanical Engineers
-
Japan Society of Mechanical Engineers
-
Society for Industrial and Applied Mathematics
-
Association for Computing Machinery
-
THE JAPAN SOCIETY OF MECHANICAL ENGINEERS
-
Association for Computing Machinery
-
Institute of Electrical and Electronics Engineers
-
THE JAPANESE SOCIETY FOR ARTIFICIAL INTELLIGENCE
Committee Memberships
-
IEEE International Conference on Cluster Computing (IEEE CLUSTER 2025) track co-chair
2025
Committee type：Academic society
-
International Conference on Machine Learning (ICML 2025) reviewer
2025
Committee type：Academic society
-
ISC High Performance (ISC 2025) program committee
2025
Committee type：Academic society
-
Platform for Advanced Scientific Computing (PASC 2025) domain co-chair
2025
Committee type：Academic society
-
39th IEEE International Parallel and Distributed Processing Symposium, (IPDPS 2025) program committee, best OSS judge
2025
Committee type：Academic society
-
International Conference on Parallel Architectures and Compilation Techniques (PACT 2025) publicity chair
2025
Committee type：Academic society
-
The International Conference for High Performance Computing, Networking, Storage and Analysis (SC 2025) proceedings vice-chair, program committee, AI4S committee
2025
Committee type：Academic society
-
Conference on Neural Information Processing Systems (NeurIPS 2025) reviewer
2025
Committee type：Academic society
-
SIAM Conference on Linear Algebra (LA24), scientific committee
2024
-
ACM International Symposium on High-Performance, Parallel and Distributed Computing (HPDC 2024), program committee
2024
-
SC Asia (SCA 2024), program committee
2024
-
Platform for Advanced Scientific Computing (PASC 2024), program committee
2024
-
The International Conference for High Performance Computing, Networking, Storage and Analysis (SC 2024), poster, ML track, AI4S/TPC WS, LLMs for HPC BoF
2024
-
International Conference on Preconditioning Techniques for Scientific and Industrial Applications, (Precond 2024), program committee
2024
-
38th IEEE International Parallel and Distributed Processing Symposium, (IPDPS 2024), track co-chair
2024
-
ISC High Performance (ISC 2024), posters chair
2024
-
The 12th International Conference on Learning Representations (ICLR 2024), reviewer
2024
-
The 30th International Conference on Parallel Processing (Euro-Par 2024), program committee
2024
-
The IEEE / CVF Computer Vision and Pattern Recognition Conference (CVPR 2024), reviewer
2024
-
38th Conference on Neural Information Processing Systems (NeurIPS 2024), reviewer
2024
-
The Eleventh International Conference on Learning Representations (ICLR 2023), reviewer
2023
Committee type：Academic society
-
Platform for Advanced Scientific Computing (PASC 2023), mini symposia and posters com- mittee, program committee
2023
-
The 23rd IEEE/ACM international Symposium on Cluster, Cloud and Internet Computing (CCGrid 2023), program committee
2023
-
The IEEE / CVF Computer Vision and Pattern Recognition Conference (CVPR 2023), re- viewer
2023
-
The 32nd International Conference on Parallel Architectures and Compilation Techniques (PACT 2023), publicity chair Asia
2023
-
IEEE International Conference on Cluster Computing (IEEE CLUSTER 2023), publicity chair
2023
-
Principles and Practice of Parallel Programming (PPoPP 2023), publicity chair
2023
-
37th IEEE International Parallel and Distributed Processing Symposium, (IPDPS 2023), pro- gram committee
2023
-
ISC High Performance (ISC 2023), posters deputy chair
2023
-
SIAM Conference on Computational Science and Engineering (CSE23), poster judge
2023
-
10th International Congress on Industrial and Applied Mathematics (ICIAM 2023), program committee
2023
-
The International Conference for High Performance Computing, Networking, Storage and Analysis (SC 2023), posters, workshops, ML tech paper, panelist, best paper
2023
-
37th Conference on Neural Information Processing Systems (NeurIPS 2023), reviewer
2023
-
Platform for Advanced Scientific Computing (PASC 2023), mini symposia and posters committee, program committee
2023
-
37th IEEE International Parallel and Distributed Processing Symposium, (IPDPS 2023), program committee
2023
-
The IEEE / CVF Computer Vision and Pattern Recognition Conference (CVPR 2023), reviewer
2023
-
IEEE CLUSTER (IEEE CLUSTER 2022), program committee
2022
Committee type：Academic society
-
ISC High Performance (ISC 2022), poster committee
2022
Committee type：Academic society
-
The 36th IEEE International Parallel & Distributed Processing Symposium (IPDPS 2022), program committee
2022
Committee type：Academic society
-
International Conference on High Performance Computing in Asia-Pacific Region (HPC Asia 2022), Applications track co-chair
2022
Committee type：Academic society
-
European Conference on Computer Vision (ECCV 2022), reviewer
2022
Committee type：Academic society
-
The International Conference for High Performance Computing, Networking, Storage and Analysis (SC 2022), program committee, workshop chair
2022
Committee type：Academic society
-
49th International Conference on Parallel Processing (ICPP 2022), program committee
2022
Committee type：Academic society
-
IEEE CLUSTER (IEEE CLUSTER 2021), program committee
2021
Committee type：Academic society
-
ISC High Performance (ISC 2021), program committee, best paper committee, steering com- mittee
2021
Committee type：Academic society
-
The 35th IEEE International Parallel & Distributed Processing Symposium (IPDPS 2021), program committee
2021
Committee type：Academic society
-
International Conference on High Performance Computing in Asia-Pacific Region (HPC Asia 2021), organizing committee
2021
Committee type：Academic society
-
SIAM Computational Science and Engineering (SIAM CSE 2021), organizing committee
2021
Committee type：Academic society
-
IEEE International Symposium on Computer Architecture and High Performance Computing (SBAC-PAD 2021), program committee
2021
Committee type：Academic society
-
48th International Conference on Parallel Processing (ICPP 2021), program committee
2021
Committee type：Academic society
-
The International Conference for High Performance Computing, Networking, Storage and Analysis (SC 2021), program committee
2021
Committee type：Academic society
-
ISC High Performance (ISC 2021), program committee, best paper committee, steering committee
2021
-
SIAM Conference on Parallel Processing for Scientific Computing (SIAM PP 2020), program committee
2020
Committee type：Academic society
-
The International Conference for High Performance Computing, Networking, Storage and Analysis (SC 2020), ML track Chair
2020
Committee type：Academic society
-
The 26th International Conference on Parallel Processing (Euro-Par 2020), program committee
2020
Committee type：Academic society
-
International Conference on High Performance Computing in Asia-Pacific Region (HPC Asia 2020), program committee
2020
Committee type：Academic society
-
48th International Conference on Parallel Processing (ICPP 2019), Applications track chair, Workshop co-organizer
2019
Committee type：Academic society
-
SIAM Conference on Computational Science and Engineering (SIAM CSE19), organizing com- mittee
2019
Committee type：Academic society
-
The International Conference for High Performance Computing, Networking, Storage and Analysis (SC 2019), program committee
2019
Committee type：Academic society
-
ISC High Performance (ISC 2019), program committee
2019
Committee type：Academic society
-
IEEE CLUSTER (IEEE CLUSTER 2019), program committee
2019
Committee type：Academic society
-
Platform for Advanced Scientific Computing Conference (PASC 2019), program committee
2019
Committee type：Academic society
-
The 3rd cross-disciplinary Workshop on Computing Systems, Infrastructures, and Program- ming (xSIG 2019), program committee
2019
Committee type：Academic society
-
IEEE 13th International Symposium on Embedded Multicore/Many-core Systems-on-Chip (MCSoC 2019), ATMG Track Chair
2019
Committee type：Academic society
-
The 3rd cross-disciplinary Workshop on Computing Systems, Infrastructures, and Programming (xSIG 2019), program committee
2019
-
SIAM Conference on Computational Science and Engineering (SIAM CSE 2019), organizing committee
2019
-
SIAM Conference on Parallel Processing for Scientific Computing (SIAM PP 2018), organizing committee
2018
Committee type：Academic society
-
The International Conference for High Performance Computing, Networking, Storage and Analysis (SC 2018), proceedings chair
2018
Committee type：Academic society
-
ISC High Performance (ISC 2018), proceedings chair
2018
Committee type：Academic society
-
IEEE CLUSTER (IEEE CLUSTER 2018), program committee
2018
Committee type：Academic society
-
The 32nd IEEE International Parallel & Distributed Processing Symposium (IPDPS 2018), program committee
2018
Committee type：Academic society
-
18th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGrid 2018), program committee
2018
Committee type：Academic society
-
SC Asia (SCA 2018), program committee
2018
Committee type：Academic society
-
The 23rd International Conference on Parallel Processing (Euro-Par 2017), program committee
2017
Committee type：Academic society
-
The International Conference for High Performance Computing, Networking, Storage and Analysis (SC 2017), program committee
2017
Committee type：Academic society
-
Platform for Advanced Scientific Computing Conference (PASC 2017), program committee
2017
Committee type：Academic society
-
The 31st IEEE International Parallel & Distributed Processing Symposium (IPDPS 2017), program committee
2017
Committee type：Academic society
-
International Conference on Supercomputing (ICS 2017), program committee
2017
Committee type：Academic society
-
IEEE CLUSTER (IEEE CLUSTER 2017), program committee
2017
Committee type：Academic society
-
The International Conference for High Performance Computing, Networking, Storage and Analysis (SC 2016), program committee
2016
Committee type：Academic society
-
16th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGrid 2016), program committee
2016
Committee type：Academic society
-
International Conference on Supercomputing (ICS 2016), program committee
2016
Committee type：Academic society
-
Platform for Advanced Scientific Computing Conference (PASC 2016), program committee
2016
Committee type：Academic society
-
ISC High Performance (ISC 2016), program committee
2016
Committee type：Academic society
-
The 30th IEEE International Parallel & Distributed Processing Symposium (IPDPS 2016), program committee
2016
Committee type：Academic society
-
The 23rd annual IEEE International Conference on High Performance Computing, Data, and Analytics (HiPC 2016), program committee
2016
Committee type：Academic society
-
28th International Symposium on Computer Architecture and High Performance Computing (SBAC-PAD 2016), program committee
2016
Committee type：Academic society
-
IEEE CLUSTER (IEEE CLUSTER 2016), program committee
2016
Committee type：Academic society
-
IEEE CLUSTER (IEEE CLUSTER 2015), program committee
2015
Committee type：Academic society
-
The International Conference for High Performance Computing, Networking, Storage and Analysis (SC 2015), program committee
2015
Committee type：Academic society
-
The 22nd annual IEEE International Conference on High Performance Computing, Data, and Analytics (HiPC 2015), program committee
2015
Committee type：Academic society
-
15th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGrid 2015), program committee
2015
Committee type：Academic society
-
14th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGrid 2014), program committee
2014
Committee type：Academic society
-
The annual IEEE International Conference on High Performance Computing (HiPC 2014), program committee
2014
Committee type：Academic society
-
The 20th International Conference on Parallel Processing (Euro-Par 2014), program committee
2014
Committee type：Academic society
-
The International Conference for High Performance Computing, Networking, Storage and Analysis (SC 2014), program committee
2014
Committee type：Academic society
-
The International Meeting on High-Performance Computing for Computational Science (VEC- PAR 2014), program committee
2014
Committee type：Academic society
-
The International Meeting on High-Performance Computing for Computational Science (VECPAR 2014), program committee
2014
-
The International Conference for High Performance Computing, Networking, Storage and Analysis (SC 2013), program committee
2013
Committee type：Academic society
-
The 28th IEEE International Parallel & Distributed Processing Symposium (IPDPS 2013), program committee
2013
Committee type：Academic society
-
18th ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming (PPoPP 2013), program committee
2013
Committee type：Academic society
-
The annual IEEE International Conference on High Performance Computing (HiPC 2013), program committee
2013
Committee type：Academic society
-
15th International Symposium on Symbolic and Numeric Algorithms for Scientific Computing (SYNASC 2013), program committee
2013
Committee type：Academic society
-
The Third International Workshop on Frontier of GPU Computing, program committee
2012
Committee type：Academic society
-
The 12th International Symposium on Parallel and Distributed Computing (ISPDC 2012), program committee
2012
Committee type：Academic society
Papers
-
Loss and Optimizer as Two Essential Mechanisms Behind Knowledge Distillation Reviewed
Satoki Ishikawa, Sameer Satish Deshmukh, Sakina Fatima, Takumi Honda, Rio Yokota
ICML Workshop HiLD 2026.7
-
Monotonic and Non-Monotonic Length-Performance Relationships in Reasoning RL Reviewed
Daisuke Nohara, Taishi Nakamura, Rio Yokota
ICML 2026 Workshop AdaptFM 2026.7
-
PowerCLIP: Powerset Alignment for Fine-Grained Contrastive Pre-Training Reviewed
Masaki Kawamura, Nakamasa Inoue, Rintaro Yanagi, Hirokatsu Kataoka, Rio Yokota
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2026.6
-
Building Effective Japanese Medical LLMs with an Open Recipe for Domain Adaptation through Continued Pre-training Reviewed
Akiko Aizawa, Yuki Arase, Fei Cheng, Jiahao Huang, Zhiyi Huang, Junfeng Jiang, Teruhito Kanazawa, Daisuke Kawahara, Kazuma Kobayashi, Takashi Kodama, Sadao Kurohashi, Yusuke Oda, Yuma Tsuta, Zhen Wan, Zhishen Yang and Rio Yokota
International Conference on Language Resources and Evaluation (LREC) 10405 - 10423 2026.5
-
Optimal Sparsity of Mixture-of-Experts Language Models for Reasoning Tasks Reviewed
Taishi Nakamura, Satoki Ishikawa, Masaki Kawamura, Takumi Okamoto, Daisuke Nohara, Jun Suzuki, Rio Yokota
International Conference on Learning Representations (ICLR) 2026.4
-
Qwen3-Swallow: Continual Pre-Training of Japanese–English Thinking LLMs with Megatron-LM Reviewed
Kazuki Fujii, Yukito Tajima, Masaki Kawamura, Sakae Mizuki, Taishi Nakamura, Hinari Shimada, Masanari Ohi, Taihei Shiotani, Koshiro Saito, Takumi Okamoto, Shigeki Ishida, Youmi Ma, Hiroya Takamura, Rio Yokota, Naoaki Okazaki
GPU Technology Conference, San Jose 2026.3
-
Swallow LLM: Continual Pre-Training and RL for Sovereign AI Reviewed
Kazuki Fujii, Rio Yokota
GPU Technology Conference (GTC) 2026.3
-
Tensor-Core-Optimized Strategies for BLR × Tall-Skinny Matrix Multiplication in BEM Reviewed
Akihiro Ida, Kazuya Goto, Rio Yokota, Tasuku Hiraishi, Toshihiro Hanawa, Takeshi Iwashita, Masatoshi Kawai, Satoshi Ohshima, Tetsuya Hoshino
Proceedings of the Supercomputing Asia and International Conference on High Performance Computing in Asia Pacific Region 153 - 164 2026.1
-
FedPM: Federated Learning Using Preconditioned Mixing of Local Parameters Reviewed
Hiro Ishii, Kenta Niwa, Hiroshi Sawada, Akinori Fujino, Noboru Harada, Rio Yokota
The 40th Annual AAAI Conference on Artificial Intelligence 2026.1
-
Satoshi Ohshima, Akihiro Ida, Masatoshi Kawai, Takeshi Fukaya, Rio Yokota
2025 IEEE 18th International Symposium on Embedded Multicore/Many-core Systems-on-Chip (MCSoC) 41 - 48 2025.12
-
Variational Learning Finds Flatter Solutions at the Edge of Stability Reviewed International journal
Rio Yokota
Annual Conference on Neural Information Processing Systems (NeurIPS) 2025.12
-
Masked Gated Linear Unit Reviewed
Yukito Tajima, Nakamasa Inoue, Yusuke Sekikawa, Ikuro Sato, Rio Yokota
Annual Conference on Neural Information Processing Systems (NeurIPS) 2025.12
-
Leveraging High-Resource English Corpora for Cross-lingual Domain Adaptation in Low-Resource Japanese Medicine via Continued Pre-training Reviewed International journal
Rio Yokota
EMNLP Findings 11469 - 11488 2025.11
-
Youmi Ma, Sakae Mizuki, Kazuki Fujii, Taishi Nakamura, Masanari Ohi, Hinari Shimada, Taihei Shiotani, Koshiro Saito, Koki Maeda, Kakeru Hattori, Takumi Okamoto, Shigeki Ishida, Rio Yokota, Hiroya Takamura, Naoaki Okazaki
Conference on Language Modeling (COLM) 2025.10
-
Optimal Sparsity of Mixture-of-Experts Language Models for Reasoning Tasks Reviewed
Taishi Nakamura, Satoki Ishikawa, Masaki Kawamura, Takumi Okamoto, Daisuke Nohara, Jun Suzuki, Rio Yokota
ICML 2025 2nd AI for Math Workshop 2025.7
-
Lion Cub: Minimizing Communication Overhead in Distributed Lion Reviewed
Satoki Ishikawa, Tal Ben-Nun, Brian Van Essen, Rio Yokota, Nikoli Dryden
ICML 2025 Workshop TTODLer-FM 2025.7
-
Scaling Large-scale GNN Training to Thousands of Processors on CPU-based Supercomputers Reviewed
Chen Zhuang, Lingqi Zhang, Du Wu, Peng Chen, Jiajun Huang, Xin Liu, Rio Yokota, Nikoli Dryden, Toshio Endo, Satoshi Matsuoka, Mohamed Wahib
Proceedings of the 39th ACM International Conference on Supercomputing 57 - 72 2025.6
-
Improving LoRA with Variational Learning.
Bai Cong, Nico Daheim, Yuesong Shen, Rio Yokota, Mohammad Emtiyaz Khan, Thomas Möllenhoff
CoRR abs/2506.14280 2025.6
-
On the Interplay Between Precision, Rank, Admissibility, and Iterative Refinement for Hierarchical Low-Rank Matrix Solvers Reviewed
Thomas Spendlhofer, Qianxiang Ma, Yasuhiro Matsumoto, Rio Yokota
ISC High Performance 2025.6
-
Rewriting Pre-Training Data Boosts LLM Performance in Math and Code. Reviewed
Kazuki Fujii, Yukito Tajima, Sakae Mizuki, Hinari Shimada, Taihei Shiotani, Koshiro Saito, Masanari Ohi, Masaki Kawamura, Taishi Nakamura, Takumi Okamoto, Shigeki Ishida, Kakeru Hattori, Youmi Ma, Hiroya Takamura, Rio Yokota, Naoaki Okazaki
CoRR abs/2505.02881 2025.5
-
Satoshi Yui, Hiromichi Kobayashi, Makoto Tsubota, Rio Yokota
Journal of the Physical Society of Japan 94 ( 4 ) 043601 2025.4
-
Satoki Ishikawa, Ryo Karakida, Rio Yokota
The 13th International Conference on Learning Representations (ICLR) 2025.4
-
Drop Upcycling: Training Sparse Mixture of Experts with Partial Re-initialization Reviewed
Taishi Nakamura, Takuya Akiba, Kazuki Fujii, Yusuke Oda, Rio Yokota, Jun Suzuki
The 13th International Conference on Learning Representations (ICLR) 2025.4
-
SuperGCN: General and Scalable Framework for GCN Training on CPU-powered Supercomputers Reviewed
Chen Zhuang, Peng Chen, Xin Liu, Nikoli Dryden, Rio Yokota, Toshio Endo, Satoshi Matsuoka, Mohamed Wahib
ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming (PPoPP) 2025.3
-
Kakeru Hattori, Sakae Mizuki, Kazuki Fujii, Taishi Nakamura, Taihei Shiotani, Kai Ueki, Takuro Niitsuma, Akira Kawabata, Hideaki Tamori, Youmi Ma, Koki Maeda, Masanari Ohi, Koshiro Saito, Takumi Okamoto, Shigeki Ishida, Rio Yokota, Hiroya Takamura, Naoaki Okazaki
2025.3
-
Youmi Ma, Sakae Mizuki, Kazuki Fujii, Taishi Nakamura, Masanari Ohi, Hinari Shimada, Taihei Shiotani, Koshiro Saito, Koki Maeda, Kakeru Hattori, Takumi Okamoto, Shigeki Ishida, Rio Yokota, Hiroya Takamura, Naoaki Okazaki
2025.3
-
Aurora-M: The First Open Source Multilingual Language Model Red-teamed according to the U.S. Executive Order Reviewed
Taishi Nakamura, Mayank Mishra, Simone Tedeschi, Yekun Chai, Jason T Stillerman, Felix Friedrich, Prateek Yadav, Tanmay Laud, Vu Minh Chien, Terry Yue Zhuo, et al.
The 31st International Conference on Computational Linguistics (COLING), Industry Track 2025.1
-
Aurora-M: Open Source Continual Pre-training for Multilingual Language and Code Reviewed
Taishi Nakamura, Mayank Mishra, Simone Tedeschi, Yekun Chai, Jason T Stillerman, Felix Friedrich, Prateek Yadav, Tanmay Laud, Vu Minh Chien, Terry Yue Zhuo, Diganta Misra, Ben Bogin, Xuan-Son Vu, Marzena Karpinska, Arnav Varma Dantuluri, Wojciech Kusa, Tommaso Furlanello, Rio Yokota, Niklas Muennighoff, Suhas Pai, Tosin Adewumi, Veronika Laippala, Xiaozhe Yao, Adalberto Junior, Alpay Ariyak, Aleksandr Drozd, Jordan Clive, Kshitij Gupta, Liangyu Chen, Qi Sun, Ken Tsui, Noah Persaud, Nour Fahmy, Tianlong Chen, Mohit Bansal, Nicolo Monti, Tai Dang, Ziyang Luo, Tien-Tung Bui, Roberto Navigli, Virendra Mehta, Matthew Blumberg, Victor May, Huu Nguyen, Sampo Pyysalo
The 31st International Conference on Computational Linguistics (COLING), Industry Track, Abu Dhabi, UAE 2025.1
-
Scaling Backwards: Minimal Synthetic Pre-Training? Reviewed
Ryo Nakamura, Ryu Tadokoro, Ryosuke Yamada, Yuki M. Asano, Iro Laina, Christian Rupprecht, Nakamasa Inoue, Rio Yokota, Hirokatsu Kataoka
European Conference on Computer Vision (ECCV) 153 - 171 2025
-
On the Relationship Between Double Descent of CNNs and Shape/Texture Bias Under Learning Process Reviewed
Shun Iwase, Shuya Takahashi, Nakamasa Inoue, Rio Yokota, Ryo Nakamura, Hirokatsu Kataoka, Eisaku Maeda
27th International Conference on Pattern Recognition (ICPR) 95 - 109 2025
-
Rethinking Image Super-Resolution from Training Data Perspectives Reviewed
Go Ohtani, Ryu Tadokoro, Ryosuke Yamada, Yuki M. Asano, Iro Laina, Christian Rupprecht, Nakamasa Inoue, Rio Yokota, Hirokatsu Kataoka, Yoshimitsu Aoki
European Conference on Computer Vision (ECCV) 19 - 36 2025
-
A Framework for Seamless Integration and Efficient Continual Pre-Training of Large Language Models Reviewed
Kazuki Fujii, Taishi Nakamura, Rio Yokota
SC’24 TPC Workshop 2024.11
-
Continual Pre-Training for Cross-Lingual LLM Adaptation: Enhancing Japanese Language Capabilities Reviewed
Kazuki Fujii, Taishi Nakamura, Mengsay Loem, Hiroki Iida, Masanari Ohi, Kakeru Hattori, Shota Hirai, Sakae Mizuki, Rio Yokota, Naoaki Okazaki
Conference on Language Modeling COLM 2024.10
-
Building a Large Japanese Web Corpus for Large Language Models Reviewed
Naoaki Okazaki, Kakeru Hattori, Shota Hirai, Hiroki Iida, Masanari Ohi, Kazuki Fujii, Taishi Nakamura, Mengsay Loem, Rio Yokota, Sakae Mizuki
Conference on Language Modeling COLM 2024.10
-
Formula-Supervised Visual-Geometric Pre-training Reviewed
Ryosuke Yamada, Kensho Hara, Hirokatsu Kataoka, Koshi Makihara, Nakamasa Inoue, Rio Yokota, Yutaka Satoh
European Conference on Computer Vision (ECCV) 57 - 74 2024.9
-
画像超解像における学習データ構築の再考 Reviewed
Go Ohtani, Ryu Tadokoro, Ryosuke Yamada, Yuki M. asano, Iro Laina, Chistian Repprecht, Nakamasa Inoue, Rio Yokota, Hirokatsu Kataoka, Yoshimitsu Aoki
2024.8
-
単眼カメラを用いたリアルタイムな３次元マップの変化検出を目的とした密なバンドル調整 Reviewed
Kai Okawa, Ken Sakurada, Rio Yokota
2024.8
-
Scaling Backwards: Minimal Synthetic Pre-training? Reviewed
Ryu Tadokoro, Ryo Nakamura, Ryosuke Yamada, Yuki M. Asano, Iro Laina, Chistian Repprecht, Nakamasa Inoue, Rio Yokota, Hirokatsu Kataoka
第27回 画像の認識・理解シンポジウム (MIRU) 2024.8
-
An Inherently Parallel H^2-ULV Factorization for Solving Dense Linear Systems on GPUs Reviewed
Qianxiang Ma, Rio Yokota
International Journal of High Performance Computing Applications 38 ( 4 ) 314 - 336 2024.7
-
Variational Learning is Effective for Large Deep Networks Reviewed
Yuesong Shen, Nico Daheim, Gian Maria Marconi, Peter Nickl, Bai Cong, Bazan Clemen, Emile Marcel Raoul, Rio Yokota, Iryna Gurevych, Daniel Cremers, Mohammad, Emtiyaz Khan, Thomas Möllenhoff
The 41st International Conference on Machine Learning (ICML) 2024.7
-
Interaction between quantum turbulence and normal-fluid turbulence in superfluid helium Reviewed
Hiromichi Kobayashi, Satoshi Yui, Makoto Tsubota, Tomokazu Saito, Rio Yokota
Thirteenth International Symposium on Turbulence and Shear Flow Phenomena (TSFP13) 2024.6
-
SIFTer: Self-improving Synthetic Datasets for Pre-training Classification Models Reviewed
Ryo Hayamizu, Shota Nakamura, Sora Takashima, Hirokatsu Kataoka, Ikuro Sato, Nakamasa Inoue, Rio Yokota
CVPR SynData Workshop 2024.6
-
When Does Second-Order Optimization Speed Up Training? Reviewed
Satoki Ishikawa, Rio Yokota
The 12th International Conference on Learning Representations (ICLR), Tiny paper 2024.5
-
DGEMM on integer matrix multiplication unit Reviewed
Hiroyuki Ootomo, Katsuhisa Ozaki, Rio Yokota
The International Journal of High Performance Computing Applications 38 ( 4 ) 297 - 313 2024.3
-
Variational Low-Rank Adaptation Using IVON. Reviewed
Bai Cong, Nico Daheim, Yuesong Shen, Daniel Cremers, Rio Yokota, Mohammad Emtiyaz Khan, Thomas Möllenhoff
CoRR abs/2411.04421 2024
-
Koshiro Saito, Sakae Mizuki, Masanari Ohi, Taishi Nakamura, Taihei Shiotani, Koki Maeda, Youmi Ma, Kakeru Hattori, Kazuki Fujii, Takumi Okamoto, Shigeki Ishida, Hiroya Takamura, Rio Yokota, Naoaki Okazaki
CoRR abs/2412.14471 2024
-
Natural Gradient Primal-Dual Method for Decentralized Learning. Reviewed
Kenta Niwa, Hiro Ishii, Hiroshi Sawada, Akinori Fujino, Noboru Harada, Rio Yokota
IEEE Trans. Signal Inf. Process. over Networks 10 417 - 433 2024
-
SegRCDB: Semantic Segmentation via Formula-Driven Supervised Learning Reviewed
Risa Shinoda, Ryo Hayamizu, Kodai Nakashima, Nakamasa Inoue, Rio Yokota, Hirokatsu Kataoka
IEEE/CVF International Conference on Computer Vision (ICCV) 19997 - 20006 2023.10
-
Pre-training Vision Transformers with Very Limited Synthesized Images Reviewed
Ryo Nakamura, Sora Takashima, Edgar Josafat Martinez-Noriega, Nakamasa Inoue, Hirokatsu Kataoka, Rio Yokota
IEEE/CVF International Conference on Computer Vision (ICCV) 20303 - 20312 2023.10
-
Cache Optimization and Performance Modeling of Batched, Small, and Rectangular Matrix Multiplication on Intel, AMD, and Fujitsu Processors Reviewed
Sameer Deshmukh, Rio Yokota, George Bosilca
ACM Transactions on Mathematical Software 49 ( 3 ) 1 - 29 2023.9
-
Computing Eigenvalue of Symmetric H^2-Matrices in Linear Time with Slicing the Spectrum Reviewed
Muhammad Ridwan Apriansyah, Rio Yokota
International Conference on Parallel Processing (ICPP) 11 - 20 2023.8
-
O(N) Distributed Direct Factorization of Structured Dense Matrices Using Runtime Systems Reviewed
Sameer Deshmukh, Rio Yokota, George Bosilca
International Conference on Parallel Processing (ICPP) 1 - 10 2023.8
-
数式ドリブン教師あり学習によるセマンティックセグメンテーション Reviewed
Risa Shinoda, Ryo Hayamizu, Kodai Nakashima, Nakamasa Inoue, Rio Yokota, Hirokatsu Kataoka
2023.7
-
学習過程における形状・テクスチャ偏重度の推移と事前学習データセットとの関係について Reviewed
Shuya Takahashi, Nakamasa Inoue, Rio Yokota, Hirokatsu Kataoka, Eisaku Maeda
2023.7
-
単眼カメラを用いたリアルタイムな3次元マップの変化検出を目的とした密なバンドル調整 Reviewed
Kai Okawa, Ken Sakurada, Rio Yokota
2023.7
-
Formula-Supervised Visual-Geometric Pre-training Reviewed
Ryosuke Yamada, Kensho Hara, Hirokatsu Kataoka, Koshi Makihara, Nakamasa Inoue, Rio Yokota, Yutaka Satoh
第26回画像の認識・理解シンポジウム (MIRU), ロングオーラル 2023.7
-
Visual Atoms: Pre-training Vision Transformers with Sinusoidal Waves Reviewed
Sora Takashima, Ryoh Hayamizu, Nakamasa Inoue, Hirokatsu Kataoka, Rio Yokota
IEEE/CVF Conference on Computer Vision and Pattern Recognition 18579 - 18588 2023.6
-
ラージバッチ学習における汎化性能の低下を抑制する正則化手法 Reviewed
Shota Nakamura, Rio Yokota
2023.6
-
受容野の自動最適化によるモードに適応的な Transformer の開発 Reviewed
Takuya Asakura, Nakamasa Inoue, Rio Yokota, Koichi Shinoda
JSAI2023 4I3OS1b05 2023.6
-
Mixed-Precision Random Projection for RandNLA on Tensor Cores Reviewed
Hiroyuki Ootomo, Rio Yokota
Platform for Advanced Scientific Computing (PASC) 1 - 11 2023.6
-
Quantum Circuit Simulation by SGEMM Emulation on Tensor Cores and Automatic Precision Selection Reviewed
Hiroyuki Ootomo, Hidetaka Manabe, Kenji Harada, Rio Yokota
ISC High Performance 259 - 276 2023.5
-
敵対的距離学習モジュールを用いた特徴変動に頑健な画像認識のための対照学習
Yoshifumi Sugiyama, Hirokatsu Kataoka, Rio Yokota, Nakamasa Inoue
2023.3
-
画像識別における形状・テクスチャ偏重度と二重降下現象の関係について Reviewed
Shuya Takahashi, Nakamasa Inoue, Rio Yokota, Hirokatsu Kataoka, Eisaku Maeda
2023.3
-
Reducing Shared Memory Footprint to Leverage High Throughput on Tensor Cores and its Flexible API Extension Library Reviewed
Hiroyuki Ootomo, Rio Yokota
HPC Asia 1 - 8 2023.2
-
Improving Continual Learning by Accurate Gradient Reconstructions of the Past Reviewed
Erik Daxberger, Siddharth Swaroop, Kazuki Osawa, Rio Yokota, Richard E. Turner, Jose Miguel Hernandez-Lobato, Mohammad Emtiyaz Khan
Transactions on Machine Learning Research 2023.2
-
Empirical Study on Optimizer Selection for Out-of-Distribution Generalization Reviewed
Hiroki Naganuma, Kartik Ahuja, Shiro Takagi, Tetsuya Motokawa, Rio Yokota, Kohta Ishikawa, Ikuro Sato, Ioannis Mitliagkas
Transactions on Machine Learning Research 2023.1
-
蒸留画像による事前学習効果についての検討 Reviewed
Ryu Tadokoro, Hirokatsu Kataoka, Rei Kawakami, Rio Yokota, Nakamasa Inoue
2022.12
-
ASDL: A Unified Interface for Gradient Preconditioning in PyTorch Reviewed
Kazuki Osawa, Satoki Ishikawa, Rio Yokota, Shigang Li, Torsten Hoefler
NeurIPS Workshop Order up! The Benefits of Higher-Order Optimization in Machine Learning 2022.12
-
Empirical Study on Optimizer Selection for Out-of-Distribution Generalization Reviewed
Hiroki Naganuma, Kartik Ahuja, Ioannis Mitliagkas, Shiro Takagi, Tetsuya Motokawa, Rio Yokota, Kohta Ishikawa, Ikuro Sato
NeurIPS Workshop Distshift abs/2211.08583 2022.12
-
QR Factorization of Block Low-Rank Matrices on Multi-Instance GPU Reviewed
Satoshi Ohshima, Akihiro Ida, Rio Yokota, Ichitaro Yamazaki
The 23rd International Conference on Parallel and Distributed Computing, Applications and Technologies (PDCAT’22) 359 - 369 2022.12
-
Informative Sample-Aware Proxy for Deep Metric Learning Reviewed
Aoyu Li, Ikuro Sato, Kohta Ishikawa, Rei Kawakami, Rio Yokota
ACM Multimedia Asia 1 - 11 2022.12
-
Scalable Linear Time Dense Direct Solver for 3-D Problems Without Trailing Sub-Matrix Dependencies Reviewed
Qianxiang Ma, Sameer Deshmukh, Rio Yokota
The International Conference for High Performance Computing, Networking, Storage, and Analysis (SC22) 1 - 12 2022.11
-
Parallel QR Factorization of Block Low-Rank Matrices Reviewed
Muhammad Ridwan Apriansyah, Rio Yokota
ACM Transactions on Mathematical Software, (2022). https://doi.org/10.1145/3538647 48 ( 3 ) 1 - 28 2022.9
-
Recovering Single Precision Accuracy from Tensor Cores While Surpassing the FP32 Theoretical Peak Performance Reviewed
Hiroyuki Ootomo, Rio Yokota
The International Journal of High Performance Computing Application 36 ( 4 ) 475 - 491 2022.7
-
Replacing Labeled Real-image Datasets with Auto-generated Contours Reviewed
Hirokatsu Kataoka, Ryo Hayamizu, Ryosuke Yamada, Kodai Nakashima, Sora Takashima, Xinyu Zhang, Edgar Josafat Martinez-Noriega, Nakamasa Inoue, Rio Yokota
IEEE/CVF Conference on Computer Vision and Pattern Recognition 21200 - 21209 2022.6
-
OPIRL: Sample Efficient Off-Policy Inverse Reinforcement Learning via Distribution Matching Reviewed
Hana Hoshino, Kei Ota, Asako Kanezaki, Rio Yokota
Proceedings of IEEE International Conference on Robotics and Automation 448 - 454 2022.5
-
Scalable and Practical Natural Gradient for Large-Scale Deep Learning Reviewed
Kazuki Osawa, Yohei Tsuji, Yuichiro Ueno, Akira Naruse, Chuan-Sheng Foo, Rio Yokota
IEEE Transactions on Pattern Analysis and Machine Intelligence 2020;PP:10.1109/TPAMI.2020.3004354 44 ( 1 ) 404 - 415 2022.1
-
RePOSE: Real-Time Iterative Rendering and Refinement for 6D Object Pose Estimation Reviewed
Shun Iwase, Xingyu Liu, Rawal Khirodkar, Rio Yokota, Kris M. Kitani
Proceedings of the International Conference on Computer Vision 2021.10
-
RePOSE: Fast 6D Object Pose Refinement via Deep Texture Rendering Reviewed
Shun Iwase, Xingyu Liu, Rawal Khirodkar, Rio Yokota, Kris M. Kitani
International Conference on Computer Vision 3283 - 3292 2021.10
-
Self-supervised Continual Pretraining for Class Incremental Image Classification Reviewed
Hikaru Nakata, Nakamasa Inoue, Rio Yokota
Proceedings CVPR CLVISION Workshop (Findings) 2021.6
-
ExaFMM: a high-performance fast multipole method library with C++ and Python interfaces Reviewed
Tingyu Wang, Rio Yokota, Lorena A. Barba
The Journal of Open Source Software, 6(61):3145 (2021). 10.21105/joss.03145 6 ( 61 ) 3145 - 3145 2021.5
-
Yuichiro Ueno, Kazuki Osawa, Yohei Tsuji, Akira Naruse, Rio Yokota
Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining 2145 - 2153 2020.8
-
Distributed Memory Task-Based Block Low Rank Direct Solver Reviewed
Sameer Deshmukh, Rio Yokota
ISC High Performance 2020 2020.6
-
Distributed Memory Task-Based Block Low Rank Direct Solver Reviewed
Sameer Deshmukh, Rio Yokota
ISC High Performance 2020 2020.6
-
Randomized SVD on TensorCores Reviewed
Hiroyuki Ootomo, Rio Yokota
ISC High Performance 2020 2020.6
-
巨大行列とAI Reviewed
横田理央
数学セミナー 59 ( 2 ) 29 - 33 2020.2
-
Effect of Mixed Precision Computing on H-Matrix Vector Multiplication in BEM Analysis Reviewed
Rise Ooi, Takeshi Iwashita, Takeshi Fukaya, Akihiro Ida, Rio Yokota
ACM International Conference Proceeding Series 92 - 101 2020.1
-
QR Decomposition of Block Low-Rank Matrices Reviewed
Muhammad Ridwan Apriansyah, Rio Yokota
HPC Asia 2020 2020.1
-
スーパーコンピューティングコンテスト2019 Reviewed
横田理央
数学セミナー 59 ( 1 ) 44 - 49 2020.1
-
Distributed Memory Task-Based Block Low Rank Direct Solver Reviewed
Sameer Deshmukh, Rio Yokota
HPC Asia 2020 (poster) 2020.1
-
Measuring the Effects to Beneficial Batch Size and Required Iteration by LARS on Neural Network Training
NAGANUMA Hiroki, TATSURO Ide, YOKOTA Rio
Proceedings of the Annual Conference of JSAI JSAI2020 4Rin169 - 4Rin169 2020
-
Regularizing the fast multipole method for use in molecular simulation Reviewed
D. S. Shamshirgar, R. Yokota, A. K. Tornberg, B. Hess
Journal of Chemical Physics 151 ( 23 ) 234113 2019.12
-
Practical deep learning with Bayesian principles Reviewed
Kazuki Osawa, Siddharth Swaroop, Anirudh Jain, Runa Eschenhagen, Richard E. Turner, Rio Yokota, Mohammad Emtiyaz Khan
Advances in Neural Information Processing Systems 32 4289 - 4301 2019.12
-
QR factorization of block low-rank matrices with weak admissibility condition Reviewed
Akihiro Ida, Hiroshi Nakashima, Tasuku Hiraishi, Ichitaro Yamazaki, Rio Yokota, Takeshi Iwashita
Journal of Information Processing 27 831 - 839 2019.11
-
TSQR on TensorCores Reviewed
Hiroyuki Ootomo, Rio Yokota
The International Conference for High Performance Computing, Networking, Storage, and Analysis 2019.11
-
On Empirical Analysis of Layer-wised Learning Rate Schedule Reviewed
Hiroki Naganuma, Rio Yokota
ACML 2019 Workshop on Statistics & Machine Learning Researchers 2019.11
-
Runtime System for GPU-based Hierarchical LU factorization Reviewed
Qianxing Ma, Rio Yokota
The International Conference for High Performance Computing, Networking, Storage, and Analysis 2019.11
-
Optimization of numerous small dense-matrix-vector multiplications in h-matrix arithmetic on gpu Reviewed
Satoshi Ohshima, Ichitaro Yamazaki, Akihiro Ida, Rio Yokota
Proceedings - 2019 IEEE 13th International Symposium on Embedded Multicore/Many-Core Systems-on-Chip, MCSoC 2019 9 - 16 2019.10
-
Distributed-memory lattice H-matrix factorization Reviewed
Ichitaro Yamazaki, Akihiro Ida, Rio Yokota, Jack Dongarra
International Journal of High Performance Computing Applications 33 ( 5 ) 1046 - 1063 2019.9
-
Tensorコアを用いたTSQR Reviewed
Hiroyuki Ootomo, Rio Yokota
2019.9
-
Yohei Tsuji, Kazuki Osawa, Yuichiro Ueno, Akira Naruse, Rio Yokota, Satoshi Matsuoka
ACM International Conference Proceeding Series ( 21 ) 21 - 8 2019.8
-
Extreme scale FMM-accelerated boundary integral equation solver for wave scattering Reviewed
Mustafa Abduljabbar, Mohammed Al Farhan, Noha Al-Harthi, Rui Chen, Rio Yokota, Hakan Bagci, David Keyes
SIAM Journal on Scientific Computing 41 ( 3 ) C245 - C268 2019.6
-
Kazuki Osawa, Yohei Tsuji, Yuichiro Ueno, Akira Naruse, Rio Yokota, Satoshi Matsuoka
Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition 2019-June 12351 - 12359 2019.6
-
Effectiveness of Smoothing for Large-batch Training Using Natural Gradient Descent Reviewed
Hiroki Naganuma, Rio Yokota
The 3rd Cross-disciplinary Workshop on Computing Systems, Infrastructures, and Programming (xSIG) 2019.5
-
A Performance Improvement Approach for Second-Order Optimization in Large Mini-batch Training. Reviewed
Hiroki Naganuma, Rio Yokota
19th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing(CCGRID) 696 - 703 2019.5
-
Exhaustive study of hierarchical allreduce patterns for large messages between GPUs Reviewed
Yuichiro Ueno, Rio Yokota
Proceedings - 19th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing, CCGrid 2019 430 - 439 2019.5
-
Effect of Mixed Precision Computing on H-Matrix Vector Multiplication in BEM Analysis. Reviewed
Rise Ooi, Takeshi Iwashita, Takeshi Fukaya, Akihiro Ida, Rio Yokota
CoRR abs/1911.00093 2019
-
Highly productive, high-performance application frameworks for Post-Petascale computing Reviewed
Naoya Maruyama, Takayuki Aoki, Kenjiro Taura, Rio Yokota, Mohamed Wahib, Motohiko Matsuda, Keisuke Fukuda, Takashi Shimokawabe, Naoyuki Onodera, Michel Müller, Shintaro Iwasaki
Advanced Software Technologies for Post-Peta Scale Computing: The Japanese Post-Peta CREST Research Project 77 - 98 2018.12
-
自然勾配近似法を用いた大規模並列深層学習におけるハイパーパラメータ最適化 Reviewed
Hiroki Naganuma, Shun Iwase, Linsho Kaku, Hikaru Nakata, Rio Yokota
2018.9
-
Performance of Hierarchical-matrix BiCGStab Solver on GPU Clusters Reviewed
Ichitaro Yamazaki, Ahmad Abdelfattah, Akihiro Ida, Satoshi Ohshima, Stanimire Tomov, Rio Yokota, Jack Dongarra
Proceedings - 2018 IEEE 32nd International Parallel and Distributed Processing Symposium, IPDPS 2018 930 - 939 2018.8
-
Fast multipole preconditioners for sparse matrices arising from elliptic equations Reviewed
Huda Ibeid, Rio Yokota, Jennifer Pestana, David Keyes
Computing and Visualization in Science 18 ( 6 ) 213 - 229 2018.3
-
Accelerating Convolutional Neural Networks Using Low Precision Arithmetic Reviewed
Hiroki Naganuma, Rio Yokota
HPC Asia 2018.1
-
Optimization of hierarchical matrix computation on GPU Reviewed
Satoshi Ohshima, Ichitaro Yamazaki, Akihiro Ida, Rio Yokota
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 10776 LNCS 274 - 292 2018
-
Verification of Low-precision Arithmetic for the Acceleration of Convolutional Neural Networks Reviewed
Hiroki Naganuma, Rio Yokota
GTC Japan 2017.12
-
Accelerating Convolutional Neural Networks Using Low-Rank Tensor Decomposition Reviewed
Kazuki Osawa, Akira Sekiya, Hiroki Naganuma, Rio Yokota
117 ( 238 ) 1 - 6 2017.10
-
Accelerating matrix multiplication in deep learning by using low-rank approximation Reviewed
Kazuki Osawa, Akira Sekiya, Hiroki Naganuma, Rio Yokota
Proceedings - 2017 International Conference on High Performance Computing and Simulation, HPCS 2017 186 - 192 2017.9
-
Acceleration of Compressed Models in Deep Learning Using Half Precision Arithmetic Reviewed
H. Naganuma, K. Osawa, A. Sekiya, R. Yokota
Japan Society for Industrial and Applied Mathematics Annual Meeting 2017.9
-
Optimization of Hierarchical Matrix Computations on a Cluster of GPUs Reviewed
Satoshi Ohshima, Ichitaro Yamazaki, Akihiro Ida, Rio Yokota
Summer United Workshops on Parallel, Distributed and Cooperative Processing 2017.7
-
Communication Reducing Algorithms for Distributed Heirarchical N-Body Methods Reviewed
Mustafa AbdulJabbar, George Markomanolis, Huda Ibeid, Rio Yokota, David Keyes
32nd International Conference, ISC High Performance 2017.6
-
Accelerating Convolutional Neural Networks Using Low-Rank Approximation Reviewed
Kazuki Oosawa, Akira Sekiya, Hiroki Naganuma, Rio Yokota
22nd Conference of Japan Computational Engineering Society 22 2017.5
-
Mustafa Abduljabbar, George Markomanolis, Huda Ibeid, Rio Yokota, David Keyes
Lecture Notes in Computer Science 10266 79 - 96 2017.2
-
Fast Multipole Method as a Matrix-Free Hierarchical Low-Rank Approximation Reviewed
Rio Yokota, Huda Ibeid, David Keyes
EIGENVALUE PROBLEMS: ALGORITHMS, SOFTWARE AND APPLICATIONS IN PETASCALE COMPUTING (EPASA 2015) 117 267 - 286 2017
-
Mustafa Abduljabbar, Mohammed Al Farhan, Rio Yokota, David Keyes
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 10417 553 - 564 2017
-
Evaluating the compression efficiency of the filters in convolutional neural networks Reviewed
Kazuki Osawa, Rio Yokota
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 10614 LNCS 459 - 466 2017
-
Tapas: An Implicitly Parallel ProgrammingFramework For Hierarchical N-body Algorithms Reviewed
Fukuda Keisuke, Maruyama Naoya, Yokota Rio, Taura Kenjiro, MATSUOKA SATOSHI
The 22nd IEEE International Conference on Parallel And Distributed Systems 1100 - 1109 2016.12
-
Communication Optimization of Distributed Memory FMM for Large Scale Boundary Element Methods Invited Reviewed
Yokota Rio
Simulation 35 ( 3 ) 147 - 153 2016.9
-
Fast Multipole Method as a Matrix-free Hierarchical Low-rank Approximation Reviewed
Rio Yokota
International Workshop on Eigenvalue Problems 2016.9
-
A Performance Model for the Communication in Fast Multipole Methods on HPC Platforms Reviewed
Huda Ibeid, Rio Yokota, David Keyes
International Journal of High Performance Computing Applications 30 ( 4 ) 423 - 437 2016.6
-
FMM と H^2(HSS) 行列のトレードオフについて Reviewed
横田理央
計算工学 21 ( 4 ) 3498 - 3501 2016
-
Scaling FMM with data-driven OpenMP tasks on multicore architectures Reviewed
Amer, A., Matsuoka, S., Pericàs, M., Maruyama, N., Taura, K., Rio Yokota, Balaji, P.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 9903 LNCS 156 - 170 2016
-
Preconditioning Sparse Matrices Using a Highly Scalable Fast Multipole Method Reviewed
Rio Yokota, Huda Ibeid, David Keyes
3rd International Workshops on Advances in Computational Mechanics 2015.10
-
Julio E. Castrillon-Candas, Marc G. Genton, Rio Yokota
Spatial Statistics 18 ( 18 ) 105 - 124 2015.4
-
N-body methods Reviewed
Mustafa AbdulJabbar, Rio Yokota
2014.11
-
Fast Multipole Preconditioners for Sparse Linear Solvers Reviewed
Huda Ibeid, Rio Yokota, Jennifer Pestana, David Keyes
Proceedings of the 11th World Congress on Computational Mechanics 2014.7
-
Communication Complexity of the Fast Multipole Method and its Algebraic Variants Invited Reviewed
Rio Yokota, George Turkiyyah, David Keyes
Supercomputing Frontiers and Innovations 1 ( 1 ) 63 - 84 2014.6
-
Petascale molecular dynamics simulation using the fast multipole method on K computer Reviewed
Ohno Yousuke, Yokota Rio, Koyama Hiroshi, Morimoto Gentaro, Hasegawa Aki, Masumoto Gen, Okimoto Noriaki, Hirano Yoshinori, Ibeid Huda, Ibeid Huda, Narumi Tetsu, Taiji Makoto
Computer Physics Communications 185 ( 10 ) 2575 - 2585 2014.6
-
Fast N-body Methods as a Compute-Bound Preconditioner for Sparse Solvers on GPUs Reviewed
YOKOTA Rio
GPU Technology Conference 2014.3
-
High Performance Numerical Algorithms for Seismic and Reservoir Simulations Reviewed
LTAIEF Hatem, YOKOTA Rio
GPU Technology Conference 2014.3
-
Fast Multipole Method Preconditioning Reviewed
PESTANA Jennifer, YOKOTA Rio, IBEID Huda, KEYES David
International Conference On Preconditioning Techniques For Scientific And Industrial Applications 2013.6
-
Investigating New Numerical Techniques for Reservoir Simulations on GPUs Reviewed
ABDELFETTAH Ahmad, LTAIEF Hatem, YOKOTA Rio
GPU Technology Conference 2013.3
-
Fork-join and data-driven execution models on multi-core architectures: Case study of the FMM Reviewed
Abdelhalim Amer, Naoya Maruyama, Miquel Pericàs, Kenjiro Taura, Rio Yokota, Satoshi Matsuoka
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 7905 255 - 266 2013
-
A Task Parallelism Meets Fast Multipole Methods Reviewed
TAURA Kenjiro, NAKASHIMA Jun, YOKOTA Rio, MARUYAMA, Naoya
Workshop on Latest Advances in Scalable Algorithms for Large-Scale Systems 2012.11
-
An FMM Based on Dual Tree Traversal for Many-core Architectures Reviewed
Rio Yokota
Journal of Algorithms and Computational Technology 7 ( 3 ) 301 - 324 2012.9
-
Petascale turbulence simulation using a highly parallel fast multipole method Reviewed
Yokota Rio, Barba Lorena, Barba Lorena, Narumi Tetsu, Yasuoka Kenji
Computer Physics Communications 184 ( 3 ) 445 - 455 2012.9
-
4096GPUを用いた4096³規模の一様等方性乱流の渦法解析
横田 理央, Barba Lorena, 成見 哲
Tsubame ESJ. : e-science journal 6 1 - 6 2012.7
-
Petascale Fast Multipole Methods on GPUs Reviewed
YOKOTA Rio
The 11th International Symposium on Parallel and Distributed Computing 2012.6
-
Data-Driven Fast Multipole Method on Distributed Memory Systems with Hardware Accelerators Reviewed
Hatem Ltaief, Rio Yokota
21st International Conference on Domain Decomposition Methods 2012.6
-
Scalable Force Directed Graph Layout Algorithms Using Fast Multipole Methods Reviewed
Enas Yunis, Rio Yokota, Aron Ahmadia
Proceedings of the 11th International Symposium on Parallel and Distributed Computing 180 - 187 2012.6
-
Recent Trends in Hierarchical N-body Methods on GPUs Reviewed
YOKOTA Rio, BARBA Lorena
GPU Technology Conference 2012.5
-
A Parallel Numerical Simulation of Dust Particles Using Direct Numerical Simulation Reviewed
Hoang Vu Nguyen, Rio Yokota, Georgiy Stenchikov
Proceedings of the European Geosciences Union General Assembly 2012.4
-
Ensemble Reviewed
Rio Yokota
(2012) 14 ( 2 ) 85 - 89 2012.4
-
Data-Driven Execution of Fast Multipole Methods Reviewed
Hatem Ltaief, Rio Yokota
Concurrency and Computation: Practice and Experience 26 ( 11 ) 1935 - 1946 2012.3
-
Optimization of Molecular Dynamics Core Program on the K computer Reviewed
Yousuke Ohno, Rio Yokota, Hiroshi Koyama, Gentaro Morimoto, Aki Hasegawa, Gen Masumoto, Tetsu Narumi, Makoto Taiji
Proceedings of JSST 2012 International Conference on Simulation Technology 2012
-
FMM Tree Construction on GPUs Reviewed
YOKOTA Rio
Ensemble 14 ( 2 ) 85 - 89 2012
-
A Task Parallel Implementation of Fast Multipole Methods Reviewed
Kenjiro Taura, Jun Nakashima, Rio Yokota, Naoya Maruyama
2012 SC COMPANION: HIGH PERFORMANCE COMPUTING, NETWORKING, STORAGE AND ANALYSIS (SCC) 617 - 625 2012
-
Scalable Fast Multipole Methods for Vortex Element Methods Reviewed
Qi Hu, Nail A. Gumerov, Rio Yokota, Lorena Barba, Ramani Duraiswami
2012 SC COMPANION: HIGH PERFORMANCE COMPUTING, NETWORKING, STORAGE AND ANALYSIS (SCC) 2012
-
Scalable Fast Multipole Methods for Vortex Element Methods Reviewed
Qi Hu, Nail A. Gumerov, Rio Yokota, Lorena Barba, Ramani Duraiswami
2012 SC COMPANION: HIGH PERFORMANCE COMPUTING, NETWORKING, STORAGE AND ANALYSIS (SCC) 1408 - + 2012
-
Petascale Turbulence Simulation Using FMM Reviewed
Rio Yokota, Tetsu Narumi, Lorena Barba, Kenji Yasuoka
IPSJ SIG Notes 2011 ( 29 ) 1 - 8 2011.11
-
Rio Yokota, L. A. Barba
Computers and Fluids 80 ( 80 ) 17 - 27 2011.10
-
Hierarchical N-body simulations with auto-tuning for heterogeneous systems Reviewed
Rio Yokota, Lorena A. Barba
Computing in Science and Engineering 14 ( 3 ) 30 - 39 2011.8
-
A Tuned and Scalable Fast Multipole Method as a Preeminent Algorithm for Exascale Systems Reviewed
Rio Yokota, Lorena Barba
International Journal of High Performance Computing Applications 26 ( 4 ) 337 - 346 2011.6
-
Parameter Tuning of a Hybrid Treecode-FMM on GPUs Reviewed
YOKOTA Rio, BARBA Lorena
The First International Workshop on Characterizing Applications for Heterogeneous Exascale Systems 2011.6
-
Comparing the Treecode with FMM on GPUs for Vortex Particle Simulations of a Leapfrogging Vortex Ring Reviewed
Rio Yokota, L. A. Barba
COMPUTERS & FLUIDS 45 ( 1 ) 155 - 161 2011.6
-
Fast Multipole Method vs. Spectral Methods for the Simulation of Isotropic Turbulence on GPUs Reviewed
Rio Yokota, Lorena Barba
Proceedings of the 23rd International Conference on Parallel Computational Fluid Dynamics 2011.5
-
Treecode and fast multipole method for N-body simulation with CUDA Reviewed
Rio Yokota, Lorena Barba
2011.2
-
Vortex methods for the simulation of turbulent flows Reviewed
Rio Yokota, Shinnosuke Obi
Journal of Fluid Science and Technology 6 ( 1 ) 14 - 29 2011.1
-
N-body Simulation and FMM on the Large-scale GPU Cluster at Nagasaki University Reviewed
YOKOTA Rio, HAMADA Tsuyoshi
Journal of the Japan Society for Computational Engineering and Science 15 ( 4 ) 2416 - 2419 2010.10
-
(Really) Fast macromolecular electrostatics – fast algorithms, open software and accelerated computing Reviewed
YOKOTA Rio, BARDHAN Jaydeep, KNEPLEY Matt, BARBA Lorena
ACS Division of Physical Chemistry 240th National Meeting 2010.8
-
Biomolecular electrostatics using a fast multipole BEM on up to 512 GPUs and a billion unknowns Reviewed
Rio Yokota, Jaydeep P. Bardhan, Matthew G. Knepley, L. A. Barba, Tsuyoshi Hamada
COMPUTER PHYSICS COMMUNICATIONS 182 ( 6 ) 1272 - 1283 2010.7
-
Performance of the Fast Multipole Method on GPUs Using Various Kernels Reviewed
Rio Yokota, Lorena Barba
Proceedings of the 9th World Congress on Computational Mechanics 2010.7
-
Comparing vortex methods and finite difference methods in a homogeneous turbulent shear flow Reviewed
R. Yokota, S. Obi
INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN FLUIDS 63 ( 7 ) 828 - 846 2010.7
-
Comparing the Treecode with FMM on GPUs for Vortex Particle Simulations of a Leapfrogging Vortex Ring Reviewed
Rio Yokota, Lorena Barba
22nd International Conference on Parallel Computational Fluid Dynamics 2010.5
-
Lagrangian simulation of turbulence using vortex methods Reviewed
YOKOTA Rio, OBI Shinnosuke
2nd International Workshops on Advances in Computational Mechanics 2010.3
-
42 TFlops Hierarchical N-body Simulation on GPUs with Applications in both Astrophysics and Turbulence Reviewed
HAMADA Tsuyoshi, YOKOTA Rio, NITADORI Keigo, NARUMI Tetsu, YASUOKA Kenji, TAIJI Makoto, OGURI Kiyoshi
Supercomputing 2009 1 - 12 2009.11
-
PetRBF--A parallel O(N) algorithm for radial basis function interpolation Reviewed
Rio Yokota, L. A. Barba, Matthew G. Knepley
COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING 199 ( 25 ) 1793 - 1804 2009.9
-
Fast Multipole Methods on GPUs for the Meshfree Simulation of Turbulence Reviewed
Rio Yokota, Tetsu Narumi, Ryuji Sakamaki, Kenji Yasuoka, Shinnosuke Obi
Proceedings of the 10th US National Congress on Computational Mechanics 2009.7
-
Fast multipole methods on a cluster of GPUs for the meshless simulation of turbulence Reviewed
Rio Yokota, Tetsu Narumi, Ryuji Sakamaki, Shun Kameoka, Shinnosuke Obi, Kenji Yasuoka
Computer Physics Communications 180 ( 11 ) 2066 - 2078 2009.6
-
DNS of Homogeneous Turbulence Using Vortex Methods Accelerated by the FMM on a Cluster of GPUs Reviewed
Rio Yokota, Tetsu Narumi, Ryuji Sakamaki, Shun Kameoka, Kenji Yasuoka, Shinnosuke Obi
Proceedings of the 21st International Conference on Parallel Compuational Fluid Dynamics 2009.5
-
Lagrangian Vortex Methods in Turbulent Channel Flows Reviewed
R. Yokota, K. Fukagata, S. Obi
ADVANCES IN TURBULENCE XII - PROCEEDINGS OF THE 12TH EUROMECH EUROPEAN TURBULENCE CONFERENCE 132 893 - 893 2009
-
2002 Numerical simulation of biological cell behavior in a micro channel by using CIP-Level Set method Reviewed
TAMURA Shuichi, YOKOTA Rio, FUKAGATA Koji
The Proceedings of The Computational Mechanics Conference 2009 ( 0 ) 546 - 547 2009
-
Validation of Vortex Methods for a Turbulent Channel Flow Reviewed
Yokota Rio, Obi Shinnosuke
2009 253 - 253 2009
-
Meshfree Simulation of Turbulence Using the Fast Multipole Methods on GPUs Reviewed
Rio Yokota, Tetsu Narumi, Ryuji Sakamaki, Shun Kameoka, Kenji Yasuoka, Shinnosuke Obi
Proceedings of the 22nd Symposium on Computational Fluid Dynamics 2008.12
-
Direct Numerical Simulation of Homogeneous Shear Flow Using Vortex Methods Reviewed
Rio Yokota, Shinnosuke Obi
Proceedings of the 4th International Conference on Vortex Flows and Vortex Models 2008.4
-
Mesh-Free Simulation of the Homogeneous Shear Flow Using Vortex Methods Reviewed
Rio Yokota, Shinnosuke Obi
Proceedings of the 23rd IIS Turbulence and Shear Flow Dynamics Symposium 2008.3
-
The Study of Colliding Vortex Rings Using a Special-Purpose Computer and FMM Reviewed
Tarun K. Sheel, Rio Yokota, Kenji Yasuoka, Shinnosuke Obi
Transactions of the Japan Society for Computational Engineering and Science, 20080003 (2008) 20080003 20080003 2008.1
-
Vortex Method Calculation of a Turbulent Channel Flow Reviewed
Yokota Rio, Fukagata Koji, Obi Shinnosuke
2008 337 - 337 2008
-
Computation of wing-tip vortex by a three-dimensional vortex method Reviewed
SATO Akira, YOKOTA Rio, OBI Shinnosuke
21st Symposium on Computational Fluid Dynamics 2007.12
-
Calculation of isotropic turbulence using a pure Lagrangian vortex method Reviewed
R. Yokota, T. K. Sheel, S. Obi
JOURNAL OF COMPUTATIONAL PHYSICS 226 ( 2 ) 1589 - 1606 2007.10
-
Pure Lagrangian vortex methods for the simulation of decaying isotropic turbulence Reviewed
YOKOTA RIO, OBI SHINNOSUKE
Proceedings of 5th Int. Symp. Turbulent Shear Flow Phenomena 365 - 370 2007.8
-
Vortex Methods for the Calculation of Homogeneous Shear Flows Reviewed
Yokota Rio, Obi Shinnosuke
2007 195 - 195 2007
-
662 Mesh-free Turbulence Simulation Using Vortex Methods Reviewed
YOKOTA Rio, OBI Shinnosuke
The Proceedings of Conference of Tokai Branch 2007 ( 0 ) 323 - 324 2007
-
Vortex flow simulation between multiple bridge decks Reviewed
YOKOTA Rio, OBI Shinnosuke
Whither Turbulence Prediction and Control 2006.3
-
2016 Simulation of a Wake using a 3-D Vortex Element Method Reviewed
YOKOTA Rio, OBI Shinnosuke
The proceedings of the JSME annual meeting 2006 ( 0 ) 31 - 32 2006
-
1202 Calculation of Fluid Structure Interaction using VEM and BEM(1) Reviewed
Yokota Rio, Obi Shinnosuke
The Proceedings of the Fluids engineering conference 2006 ( 0 ) _1202 - a_ 2006
-
複数の鈍い形状物体周りの渦流れシミュレーション Reviewed
Rio Yokota, Shinnosuke Obi
2005.12
-
Vortex flow simulation of multiple bluff bodies
YOKOTA Rio, TOKAI Norihiko, OBI Shinnosuke
The Proceedings of The Computational Mechanics Conference 2004 ( 0 ) 731 - 732 2004
Books
-
巨大行列とAI
Rio Yokota（ Role： Contributor）
2020.2
-
スーパーコンピューティングコンテスト2019
Rio Yokota（ Role： Contributor）
2020.1
-
High Performance Parallelism Pearls
YOKOTA Rio（ Role： Joint editorN-body methods on Xeon Phi coprocessors）
Morgan Kaufmann 2014.11 （ ISBN:9780128021187 ）
-
GPU Computing Gems
Rio Yokota, Lorena Barba（ Role： ContributorTreecode and fast multipole method for N-body simulation with CUDA）
Morgan Kaufmann 2011 （ ISBN:0123849888 ）
MISC
-
大規模言語モデルの強化学習による効率的な論理推論
Daisuke Nohara, Taishi Nakamura, Rio Yokota
2026.3
-
視覚・言語モデルにおける日本語OCR性能の向上
Shungo Yasuda, Rio Yokota
2026.3
-
One-Shot NASによるBERTのモデル圧縮
Takumi Okamoto, Rio Yokota
2024.5
-
Kazuki Fujii, Taishi Nakamura, Mengsay Loem, Hiroki Iida, Masanari Ohi, Kakeru Hattori, Shota Hirai, Sakae Mizuki, Rio Yokota, Naoaki Okazaki
2024.3
-
継続学習を用いた効率の良いマルチリンガル・マルチエキスパートモデルの開発
Taishi Nakamura, Rio Yokota
2024.3
-
大規模言語モデルの日本語能力の効率的な強化: 継続事前学習における語彙拡張と対訳コーパスの活用
Sakae Mizuki, Hiroki Iida, Kazuki Fujii, Taishi Nakamura, Mengsay Loem, Masanari Ohi, Kakeru Hattori, Shota Hirai, Rio Yokota, Naoaki Okazaki
2024.3
-
Naoaki Okazaki, Kakeru Hattori, Shota Hirai, Hiroki Iida, Masanari Ohi, Kazuki Fujii, Taishi Nakamura, Mengsay Loem, Rio Yokota, Sakae Mizuki
2024.3
-
Takumi Okamoto, Rio Yokota
2024.3
-
大規模言語モデルの分散並列学習
Kazuki Fujii, Rio Yokota
2024.3
-
Visual SLAM を目的とした深度の一貫性を考慮した密なバンドル調整
Kai Okawa, Ken Sakurada, Rio Yokota
2024.1
-
Attempt to improve the computational performance by multi-processes GPU execution
大島聡史, 伊田明弘, 河合直聡, 深谷猛, 横田理央, 山崎市太郎
計算工学講演会論文集(CD-ROM) 29 2024
-
超解像事前学習における核心的要素の解明
Go Ohtani, Ryu Tadokoro, Hirokatsu Kataoka, Nakamasa Inoue, Rio Yokota, Yoshimitsu Aoki
2023.12
-
Two-way coupled simulation of quantum turbulence and normal-fluid turbulence in superfluid helium-4
Hiromichi Kobayashi, Satoshi Yui, Makoto Tsubota, Tomokazu Saito, Rio Yokota
APS DFD 76th Annual Meeting of the Division of Fluid Dynamics 2023.11
-
Two-phase flow of quantum turbulence and normal-fluid turbulence in superfluid helium-4
Hiromichi Kobayashi, Satoshi Yui, Makoto Tsubota, Rio Yokota
ERCOFTAC Symposium on Engineering Turbulence Modeling and Measurements (ETMM) 2023.9
-
超流動ヘリウム4の量子乱流:常流体乱流による渦糸バンドルの形成
Satoshi Yui, Hiromichi Kobayashi, Makoto Tsubota, Tomokazu Saito, Rio Yokota
2023.9
-
Vortex-Filament Bundle Induced by Normal-Fluid Turbulence in Turbulent Superfluid Helium-4
Satoshi Yui, Hiromichi Kobayashi, Makoto Tsubota, Tomokazu Saito, Rio Yokota
International Symposium on Quantum Fluids and Solids (QFS) 2023.8
-
Vortex-filament bundle induced by normal-fluid turbulence in turbulent superfluid helium-4
Hiromichi Kobayashi, Satoshi Yui, Makoto Tsubota, Tomokazu Saito, Rio Yokota
International Symposium on Quantum Fluids and Solids (QFS) 2023.8
-
Tensorコアによる単精度行列積エミュレーションの自動精度選択を用いた量子回路シミュレーション
Hiroyuki Ootomo, Hidetaka Manabe, Kenji Harada, Rio Yokota
2023.3
-
GPUとA64FXにおけるTransformerの性能比較
Shukai Nakamura, Rio Yokota
2023.3
-
ニュートンフラクタル画像による事前学習効果
Toshiki Omi, Ryo Nakamura, Hirokatsu Kataoka, Nakamasa Inoue, Rio Yokota
2023.3
-
深層学習における勾配の前処理法に関する検討
Satoki Ishikawa, Rio Yokota
2023.3
-
量子渦計算の高速多重極展開法を用いた高速化
Tomokazu Saito, Rio Yokota
2023.3
-
O(N) Factorization of Dense Matrices on GPUs Without Trailing Submatrix Dependencies
Qianxiang Ma, Rio Yokota
SIAM Conference on Computational Science and Engineering (CSE) 2023.2
-
Parallel QR Factorization of Block Low-Rank Matrices
Muhammad Ridwan Apriansyah, Rio Yokota
SIAM Conference on Computational Science and Engineering (CSE) 2023.2
-
CUDA Fortran+MIG+UVMを用いたBLR行列QR分解の大規模高速化
Satoshi Ohshima, Akihiro Ida, Masatoshi Kawai, Rio Yokota, Ichitaro Yamazaki
2023 ( HPC-190 ) 2023
-
対称ブロック低ランク行列の精度保証付き固有値問題解法
Akihiro Ida, Takeshi Ogita, Rio Yokota
2022.9
-
Tensorコアを用いた単精度行列積エミュレーションのアプリケーションでの評価
Hiroyuki Ootomo, Rio Yokota
2022.7
-
走行動画の大規模自己教師あり学習の検討と計画
Tomoya Takahashi, Shingo Yashima, Kohta Ishikawa, Ikuro Sato, Rio Yokota
2022.7
-
ViTのファインチューニング時におけるNASのモデル縮小効果
Xinyu Zhang, Sora Takashima, Rio Yokota
2022.6
-
Vision Transformerにおけるバッチサイズの汎化性能への影響
Shukai Nakamura, Rio Yokota
2022.3
-
深層学習における2次最適化の汎化性能の検証
Hiro Ishii, Rio Yokota
2022.3
-
マルチインスタンスGPU上におけるBLR行列のQR分解
大島聡史, 伊田明弘, 横田理央, 山崎市太郎
日本応用数理学会年会講演予稿集(CD-ROM) 2022 2022
-
TensorCoreを用いた精度補正単精度行列積
Hiroyuki Ootomo, Rio Yokota
2021.7
-
画像分類のための継続的な事前学習における教師なし表現学習の堅牢性に関する検証
Hikaru Nakata, Rio Yokota
2020.6
-
TensorコアのAPIの構造解析を用いた拡張ライブラリの開発
Hiroyuki Ootomo, Rio Yokota
2020.3
-
Takahiro Shohata, Hiroki Naganuma, Rio Yokota
2020 ( 1 ) 359 - 360 2020.2
-
早期終了タイミングを予測する：深層学習における確率勾配の分布の変化点検出
Keita Yashima, Kohta Ishikawa, Ikuro Sato, Tetsuhiro Nomura, Rio Yokota, Satoshi Matsuoka
2019.11
-
Flexible and Simplistic Hierarchical Matrix-Based Fast Direct Solver
P. Spalthoff, R. Yokota
The 170th Workshop on High Performance Computing 2019.7
-
GPU Implementation of TSQR Using Tensor Cores
H. Ootomo, R. Yokota
The 170th Workshop on High Performance Computing 2019.7
-
Improving the Generalization Gap in Large-batch Training Using Noise Injection
Hiroki Naganuma, Rio Yokota
IEICE General Conference 2019.3
-
Second Order Optimization for Large Scale Parallel Deep Learning
Rio Yokota, Kazuki Osawa, Yohei Tsuji, Yuichiro Ueno, Akira Naruse
IEICE General Conference 2019.3
-
Batched QR Decomposition Using TensorCores
Hiroyuki Ootomo, Rio Yokota
The 81st National Convention of IPSJ 2019.3
-
Variational Inference in Deep Learning Using Natural Gradient Descent
Hikaru Nakata, Kazuki Osawa, Rio Yokota
The 81st National Convention of IPSJ 2019.3
-
Second Order Optimization for Distributed Data-parallel Deep Learning on 4000 GPUs
Rio Yokota, Yohei Tsuji, Kazuki Osawa
I2R-TokyoTech Co-workshop on DL 2.0 2019.3
-
Fisher情報行列の解析に基づく大規模深層学習のための二次最適化手法
Kazuki Osawa, Rio Yokota, Chuan-Sheng Foo, Vijay Chandrasekhar
2019.3
-
Hiroki Naganuma, Rio Yokota
2019 ( 1 ) 315 - 316 2019.2
-
GPUによる階層型行列計算法の高速化に向けた多数の小密行列ベクトル積計算の最適化
Satoshi Ohshima, Ichitaro Yamazaki, Akihiro Ida, Rio Yokota
2019 2019
-
K-FACと分散処理による深層学習の高速化
Yohei Tsuji, Kazuki Osawa, Rio Yokota, Satoshi Matsuoka
2018.7
-
Software Auto-Tuning for Hierarchical Matrix Computation
大島 聡史, 山崎 市太郎, 伊田 明弘, 横田 理央
計算工学講演会論文集 Proceedings of the Conference on Computational Engineering and Science 23 2018.6
-
Deep Learning Using Kronecker-factored Approximation of Fisher Matrix
Hiroyuki Ohtomo, Kazuki Osawa, Rio yokota
The 80th National Convention of IPSJ 2018.3
-
Hyper-parameter Tuning for Approximate Natural Gradient Methods
Yuji Kuwamura, Kazuki Osawa, Rio Yokota
The 80th National Convention of IPSJ 2018.3
-
Distributed Learning of Deep Neural Networks Using the Kronecker Factorization of the Fisher Information Matrix
Hiroyuki Otomo, Kazuki Osawa, Rio Yokota
The 163rd Workshop on High Performance Computing 2018.3
-
Preface
Rio Yokota, Weigang Wu
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 10776 LNCS V - VI 2018
-
Preface
Rio Yokota, Michèle Weiland, David Keyes, Carsten Trinitis
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 10876 LNCS V - VI 2018
-
Preface
Rio Yokota, Michèle Weiland, John Shalf, Sadaf Alam
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 11203 LNCS V 2018
-
深層学習における低精度演算を用いた高速化及びアクセラレーターの性能評価
Hiroki Naganuma, Akira Sekiya, Kazuki Osawa, Hiroyuki Ootomo, Yuji Kuwamura, Rio Yokota
2017.10
-
Using Low-Rank Approximation in Convolutional Neural Networks
Yoshifumi Motoyama, Toshio Endo, SATOSHI MATSUOKA, Rio Yokota, Keisuke Fukuda
158th Research Presentation Seminar in High Performance Computing 2017-HPC-158 ( 25 ) 2017.3
-
階層型行列計算のGPU向け最適化
大島聡史, 山崎市太郎, 伊田明弘, 横田理央
日本応用数理学会年会講演予稿集(CD-ROM) 2017 2017
-
A Matrix-free Preconditioner for the Helmholtz Equation based on the Fast Multipole Method
Huda Ibeid, Rio Yokota, David Keyes
2016.8
-
A Matrix-Free Preconditioner for Elliptic Solvers Based on the Fast Multipole Method
Huda Ibeid, Rio Yokota, David Keyes
SIAM Conference on Parallel Processing for Scientific Computing 2016.4
-
Comparison of FMM and HSS at Large Scale
YOKOTA Rio, ROUET, Francois-Henri, LI Xiaoye
SIAM Conference on Applied Linear Algebra 2015.10
-
Fast Multipole Method as Preconditioner
Huda Ibeid, Jennifer Pestana, Rio Yokota, David Keyes
SIAM Conference on Computational Science and Engineering 2015.3
-
Asynchronous Execution of the Fast Multipole Method Using Charm++
Mustafa AbdulJabbar, Rio Yokota, David Keyes
arXiv:1405.7487 2014.5
-
Fast Multipole Method as a Preconditioner
IBEID Huda, YOKOTA Rio, KEYES David
SIAM Conference on Computational Science and Engineering 2013.2
-
Towards a Dataflow FMM using the OmpSs Programming Model
Miquel Pericas, Abdelhalim Amer, Keisuke Fukuda, Naoya Maruyama, Rio Yokota, Satoshi Matsuoka
研究報告ハイパフォーマンスコンピューティング(HPC) 2012 ( 12 ) 1 - 7 2012.9
-
Parallelizing ExaFMM with MassiveThreads Task Parallel Library and Its Evaluation
Kenjiro Taura, Jun Nakashima, Rio Yokota, Naoya Maruyama
2012 ( 13 ) 1 - 13 2012.7
-
Turbulence Simulation Using 4096³ Vortex Particles on 4096 GPUs
Yokota Rio, Barba Lorena, Narumi Tetsu
Tsubame ESJ. : e-science journal 6 17 - 22 2012.7
-
Scaling Fast Multipole Methods up to 4000 GPUs
YOKOTA Rio, NARUMI Tetsu, BARBA Lorena, YASUOKA Kenji
ATIP/A*CRC Workshop on Accelerator Technologies for High Performance Computing 2012.5
-
Mesh-free direct numerical simulation of turbulence using the vortex method on parallel MDGRAPE-3 boards along with the fast multipole method
YOKOTA Rio, NARUMI Tetsu, YASUOKA Kenji, EBISUZAKI Toshikazu, OBI Shinnosuke
Next-Generation Supercomputing Symposium 2007.10
Presentations
-
Matrices in Deep Neural Networks and How to Compute Them in Parallel Invited International conference
Rio Yokota
IEEE CLUSTER 2022 2022.9
-
Fast N-body Methods on Many-core and Heterogenous Systems International conference
YOKOTA Rio
International Workshop on Computational Science and Numerical Analysis 2012.3
-
Petaflops Scale Turbulence Simulation on TSUBAME 2.0 International conference
YOKOTA Rio
GPU@BU Workshop 2011.11
-
Large Scale Multi-GPU FMM for Bioelectrostatics International conference
YOKOTA Rio, BARBA Lorena
Presentations SIAM Conference on Computational Science and Engineering 2011.2
-
12 Steps to a Fast Multipole Method on GPUs International conference
YOKOTA Rio
Pan-American Advanced Studies Institute 2011.1
-
RBF interpolation using Gaussians with domain decomposition on GPUs International conference
YOKOTA Rio, BARBA Lorena
SIAM annual meeting 2010.7
-
Calculation of the decay of colliding turbulent vortex rings
4th International Conference on Vortex Flows and Vortex Models 2008
-
Simulation of a wake using 3-D vortex element method
Annual Meeting of the JSME 2006
-
Vortex flow simulation between multipole bridge decks
Whither Turbulence Prediction and Control 2006
-
Simulation of homogeneous isotropic turbulence using the vortex method,
YOKOTA Rio, OBI Shinnosuke
20th Symposium on Computational Fluid Dynamics 2006.12
-
Range of Applications for the Fast Multipole Method on GPUs International conference
YOKOTA Rio
Accelerated Computing 2010.1
-
Effect of Mixed Precision Computing on H-Matrix Vector Multiplication in BEM Analysis
Rise Ooi, Takeshi Iwashita, Takeshi Fukaya, Akihiro Ida, Rio Yokota
HPC Asia 2020 2020.1
-
Advances in Fast Multipole Methods for Scalable Electrostatics Calculations Invited International conference
YOKOTA Rio
Workshop: Electrostatics methods in Molecular Simulation 2013.5
-
ExaFMM – a Testbed for Comparing Various Implementations of the FMM International conference
YOKOTA Rio
SIAM Conference on Computational Science and Engineering 2015.3
-
Communication Complexity of the Fast Multipole Method and its Alge- braic Variants International conference
YOKOTA Rio, KEYES David
CBMS-NSF Conference: Fast Direct Solvers for Elliptic PDEs 2014.6
-
Various Implementations of FMM and Their Performance on Future Architectures International conference
YOKOTA Rio
Multi-resolution Interactions Workshop 2015.8
-
Running Fast Multipole Method on the Full Node of TSUBAME and K computer International conference
YOKOTA Rio
Scalable Hierarchical Algorithms for Extreme Computing 2012.4
-
Petascale Fast Multipole Methods on GPUs Invited International conference
YOKOTA Rio
The 11th International Symposium on Parallel and Distributed Computing 2012.6
-
Compute-Memory Tradeoff in Hierarchical Low-Rank Approximation Methods International conference
Rio Yokota
SIAM Conference on Computational Science and Engineering 2017.2
-
Energy Conservation of Fast Multipole Methods in Classical Molecular Dynamics Simulations Invited International conference
Rio Yokota
7th AICS International Symposium 2017.2
-
Hierarchical Low-Rank Approximations at Extreme Scale Invited International conference
Rio Yokota
32nd International Conference, ISC High Performance 2017.6
-
A Common API for Fast Multipole Methods International conference
YOKOTA Rio
Accelerate Data Analytics and Computing Workshop 2016.1
-
Tuning Parameters in FMM Invited
YOKOTA Rio
Seventh Symposium on Automatic Tuning Technology and its Application 2015.12
-
Improving Data Locality of Fast Multipole Methods International conference
YOKOTA Rio
Third Workshop on Programming Abstractions for Data Locality 2016.10
-
Fast Multipole Method Library for Multiple Architectures and its Application to Molecular and Fluid Simulations
YOKOTA Rio
8th Symposium of the Joint Usage/Research Center for Interdisciplinary Large-scale Information Infrastructures 2016.7
-
Perforamance Portability of FMM
YOKOTA Rio
21st Conference of Japan Computational Engineering Society 2016.5
-
O(N)で並列性の高い密行列のLU分解 Invited
Rio Yokota
2022.12
-
人工画像を用いたVision Transformerの大規模事前学習 Invited
Rio Yokota
2022.9
-
人工画像を用いたVision Transformerの大規模事前学習 Invited
Rio Yokota
2022.7
-
Scaling Deep Learning to Thousands of GPUs Invited International conference
Rio Yokota
HPC 2018 2018.7
-
Energy Conserving Fast Multipole Methods for the Calculation of Long-range Interactions Invited International conference
Rio Yokota
Mathematics in Action: Modeling and analysis in molecular biology and electro- physiology 2018.6
-
Optimization Methods for Large Scale Distributed Deep Learning Invited International conference
Rio Yokota
IPAM Workshop I: Big Data Meets Large-Scale Computing 2018.9
-
Early Application Results on TSUBAME 3 Invited International conference
Rio Yokota
Smoky Mountains Computational Sciences and Engineering Conference 2018.8
-
Can we use Hierarchical Low-Rank Approximation for Deep Learning? Invited International conference
Rio Yokota
HPC Saudi 2018 2018.3
-
Recent Trends in Hierarchical Low-Rank Approximation Methods Invited International conference
Rio Yokota
Tokyo Institute of Technology and Stony Brook University Joint Science and Technology Meeting 2019.5
-
階層的低ランク近似法の最新研究動向と応用例 Invited
Rio Yokota
2019.5
-
ImageNetベンチマークの大規模並列深層学習 Invited
Rio Yokota
2019.7
-
深層学習の高速化と大規模並列化 Invited
Rio Yokota
2019.9
-
深層学習に現れる密行列も構造 Invited
Rio Yokota
2019.10
-
近似行列分解と分散深層学習 Invited
Rio Yokota
2019.11
-
Degree of Approximation and Overhead of Computing Curvature, Information, and Noise Matrices Invited International conference
Rio Yokota
ICML Workshop "Beyond first order methods in machine learning systems", July 2020
-
FMMの自動チューニング可能なパラメータについて Invited
Rio Yokota
2015.12
-
Portability of the Performance of FMM Invited
Rio Yokota
2016.5
-
Kronecker Factorization for Second Order Optimization in Deep Learning Invited International conference
Rio Yokota
SIAM CSE 2019.2
-
Overview of Distributed Memory Parallelism in Deep Learning Invited International conference
Rio Yokota
DD26 MS01: Learning, Algorithms, Domain Decomposition Methods, and Applications 2020.12
-
スパコンを用いた大規模並列分散深層学習 Invited
Rio Yokota
2021.3
-
Overview of structured low-rank approximation methods Invited International conference
Rio Yokota
IUTAM Symposium on Computational Methods for Large-Scale and Complex Wave Problems 2021.6
-
階層的低ランク近似法に関するレビュー Invited
Rio Yokota
2021.9
-
Approximations of Natural Gradient Descent in Distributed Training Invited International conference
Rio Yokota
INFORMS Annual Meeting Session: Beyond first order methods in machine learning systems I 2021.10
-
大規模並列深層学習の基礎 Invited
Rio Yokota
2021.10
-
二次最適化を用いたImageNetの大規模分散深層学習 Invited
Rio Yokota
2020.1
-
二次最適化を用いた巨大な言語モデルの学習およびFRNNを用いたプラズマ挙動予測 Invited
Rio Yokota
2020.2
-
Distributed Deep Learning with Second Order Information Invited International conference
Rio Yokota
SPCL_Bcast(COMM_WORLD) 2020.10
-
深層学習におけるヘッセ行列, フィッシャー行列, 共分散行列の高速近似解法 Invited
Rio Yokota
2020.10
Awards
-
ACM Gordon Bell Prize (price/performance)
2009
Research Projects
-
高性能都市モデル計算
2026 - 2029
日本学術振興会 科学研究費助成事業 基盤研究(B)
-
次世代計算機の潜在能力を引き出すための科学技術ソフトウェアの刷新
Grant number：25H01109 2025.4 - 2028.3
日本学術振興会 科学研究費助成事業 基盤研究(A)
横田 理央
Grant amount：\46020000 （ Direct Cost: \35400000 、 Indirect Cost：\10620000 ）
-
GPU教育教材の作成及び教育プログラムの実施、生成AI技術を中心としたGPU計算資源の高度活用促進
2025 - 2030
文部科学省 文部科学省 次世代HPC・AI開発支援拠点事業
-
大規模言語モデル（LLM：Large Language Model）を活用した医薬品等の有効性・安全 性評価のためのアウトカム抽出の方法論の確立に向けた研究
Grant number：24AC0401 2024.4 - 2027.3
厚生労働省 厚生労働科学研究費 応用研究
武藤 学, 松本 繁巳, 中島 貴子, 黒田 知宏, 吉原 博幸, 小林 慎治, 粂 直人, 横田 理央, 加藤 康之
-
低ランク構造行列法の適用範囲拡大と多様な計算アーキテクチャの活用
Grant number：24K02949 2024.4 - 2027.3
日本学術振興会 科学研究費助成事業 基盤研究(B)
伊田 明弘, 横田 理央, 塙 敏博, 岩下 武史, 大島 聡史, 星野 哲也, 平石 拓, 河合 直聡
Grant amount：\18590000 （ Direct Cost: \14300000 、 Indirect Cost：\4290000 ）
-
大規模言語モデルのミスアライメントに対するレッドチーミング基盤
2024 - 2030
科学技術振興機構 JST K-program
-
大規模言語モデルの分散並列事前学習の実施に関する研究
2024 - 2029
科学技術振興機構 JST次世代人工知能技術等研究開発拠点形成事業
-
人工生成データ及び少量実データの継続事前学習による限定資源下でのAI基盤モデル構築
2024 - 2028
日本学術振興会 国際共同研究加速基金 海外連携研究
-
Next-generation high-performance linear solver for future computational science and engineering
Grant number：23H00462 2023.4 - 2027.3
Japan Society for the Promotion of Science Grants-in-Aid for Scientific Research Grant-in-Aid for Scientific Research (A)
Grant amount：\45500000 （ Direct Cost: \35000000 、 Indirect Cost：\10500000 ）
-
深層生成モデルを活用した構成的なパターン認識・理解
Grant number：23H00490 2023.4 - 2026.3
日本学術振興会 科学研究費助成事業 基盤研究(A)
篠田 浩一, 井上 中順, 横田 理央, 川上 玲, 佐藤 育郎
Grant amount：\47190000 （ Direct Cost: \36300000 、 Indirect Cost：\10890000 ）
本研究課題では，識別の対象（インスタンス）を属性の集合（束）とみなし，特徴量空間においてその特徴を属性ごとに分解する．そして，これらの属性特徴からインスタンスを再合成する過程で属性特徴を最適化することで，各属性を高精度で識別し，かつ，外れ値に対し頑健な識別手法を実現することを目的としている。このために深層生成モデルと高密度な属性アノテーションに基づく学習手法を開発する．従来研究の多くが対象とその属性が一対一に対応する平坦な意味構造を仮定していたのに対し，本研究は多くの属性が複雑に絡み合う対象における複数の属性を同時に識別することを可能にする．新しい属性やクラスの創発も視野に入れる．より具体的には、深層学習を用いた「合成による識別」のアプローチにより，構成的なパターン認識・理解を行う方法論を確立する．人の動作認識，話者・感情認識，マルチモーダル認識の3つのタスクで横断的に評価し，従来に比べ高い識別性能を目指す．初年度である本年度は、人の動作認識、話者・感情認識、マルチモーダル認識の各々の課題において、評価データベースの構築と、ベースライン方式の開発を行った。これらと並行して、比較的小規模なタスクで、拡散モデルなどの生成モデルを用いて識別を行う方式の開発を行った。また、ニューラル構造探索などを用いて生成モデルの効率的な学習を行う方式も開発した。特に、センサーと映像のマルチモーダル認識における基本方式の構築、およびデータベース構築、人間の歩容認識の基本方式の開発、マルチモーダル感情認識の基本方式の開発を行った。
-
Functional improvement of mixing promotion in fluid equipment by elucidating the universal statistical law of two-phase turbulence
Grant number：22H01403 2022.4 - 2026.3
Japan Society for the Promotion of Science Grants-in-Aid for Scientific Research Grant-in-Aid for Scientific Research (B)
Grant amount：\17290000 （ Direct Cost: \13300000 、 Indirect Cost：\3990000 ）
-
Functional improvement of mixing promotion in fluid equipment by elucidating the universal statistical law of two-phase turbulence
Grant number：23K22674 2022.4 - 2026.3
Japan Society for the Promotion of Science Grants-in-Aid for Scientific Research Grant-in-Aid for Scientific Research (B)
Grant amount：\17290000 （ Direct Cost: \13300000 、 Indirect Cost：\3990000 ）
-
Fast and Accurate Eigenvalue Calculations by Hierarchical Low-rank Approximation and its Application to Large-scale Electronic Structure Calculations
Grant number：22H03598 2022.4 - 2025.3
Japan Society for the Promotion of Science Grants-in-Aid for Scientific Research Grant-in-Aid for Scientific Research (B) Grant-in-Aid for Scientific Research (B)
Rio Yokota, Akihiro Ida, Takeshi Ogita, Takeo Hoshi
Authorship：Principal investigator
Grant amount：\17680000 （ Direct Cost: \13600000 、 Indirect Cost：\4080000 ）
-
Fast and accurate eigenvalue calculations by hierarchical low-rank approximation and its application to large-scale electronic structure calculations
Grant number：23K24854 2022.4 - 2025.3
Japan Society for the Promotion of Science Grants-in-Aid for Scientific Research Grant-in-Aid for Scientific Research (B)
Grant amount：\17680000 （ Direct Cost: \13600000 、 Indirect Cost：\4080000 ）
-
A New Bayes Duality Principle for Adaptive, Robust, Life-long Learning of AI
Grant number：JY210177nn 2021.10 - 2027.3
JST CREST
Emtiyaz Khan, Kenichi Bannai, Rio Yokota, Julyan Arbel
Authorship：Coinvestigator(s)
-
Construction of numerical linear algebra based on lattice H-matrices and its high-performance implementation on modern architectures
Grant number：21H03447 2021.4 - 2024.3
Japan Society for the Promotion of Science Grants-in-Aid for Scientific Research Grant-in-Aid for Scientific Research (B)
Grant amount：\17290000 （ Direct Cost: \13300000 、 Indirect Cost：\3990000 ）
-
Application of Unconventional Linear Algebra Techniques to Continuous Learning in Supergiant Neural Networks
Grant number：20K20624 2020.7 - 2023.3
Japan Society for the Promotion of Science Grants-in-Aid for Scientific Research Grant-in-Aid for Challenging Research (Pioneering) Grant-in-Aid for Challenging Research (Pioneering)
Grant amount：\25350000 （ Direct Cost: \19500000 、 Indirect Cost：\5850000 ）
-
Life-Long Deep Learning using Bayesian Principles
Grant number：20H04247 2020.4 - 2023.3
Japan Society for the Promotion of Science Grants-in-Aid for Scientific Research Grant-in-Aid for Scientific Research (B) Grant-in-Aid for Scientific Research (B)
Grant amount：\18200000 （ Direct Cost: \14000000 、 Indirect Cost：\4200000 ）
-
Linear Solvers for Machine Learning Hardware
Grant number：18H03248 2018.4 - 2021.3
Japan Society for the Promotion of Science Grants-in-Aid for Scientific Research Grant-in-Aid for Scientific Research (B) Grant-in-Aid for Scientific Research (B)
Yokota Rio
Grant amount：\16900000 （ Direct Cost: \13000000 、 Indirect Cost：\3900000 ）
The trend in computer architecture has now shifted from general purpose accelerators to specialized hardware for machine learning. The present work focuses on the affinity between hierarchical low-rank approximation methods, and low-precision arithmetic units and tensor product accelerators in machine learning processors to develop a suitable linear algebra library for future architectures. In FY2018, we ported our H-matrix library to use batched MAGMA operations in order to take advantage of the tensor product accelerators. In FY2019, we optimized the inner kernels of the H-matrix by making use of TensorCores. In FY2020, we extended this work to recover the accuracy when using TensorCores and measured the energy efficiency.
-
Enhancement of H-matrix library and optimization for next generation supercomputers
Grant number：17H01749 2017.4 - 2020.3
Japan Society for the Promotion of Science Grants-in-Aid for Scientific Research Grant-in-Aid for Scientific Research (B) Grant-in-Aid for Scientific Research (B)
Ida Akihiro
Grant amount：\18850000 （ Direct Cost: \14500000 、 Indirect Cost：\4350000 ）
In this study, we enhanced the HACApK, which is a library for H-matrices: the dynamic load balancing technique is introduced for H-matrix generation, and algorithms of H-matrix-vector multiplication for GPU computing and mixed-precision computing are developed and implemented. These new implementations are several to ten times faster than existing HACApK. We proposed a novel variant of low-rank structured matrices, called “lattice H-matrices”, which allow the construction of efficient operation and communication patterns compared to the conventional H-matrices. In numerical experiments for performing H-matrix-vector multiplications, the lattice H-matrices is several tens of times faster than the normal H-matrices when several thousand processes are used. Moreover, we developed an LU decomposition method based on the lattice H-matrices, and a QR decomposition method for the BLR matrices which is a simple version of lattice H-matrices. We also proposed their parallelization algorithms.
-
Acceleration and economization of deep learning algorithms for image processing in social infrastructure
2016.11 - 2022.3
Japan Science and Technology Agency CREST
SHINODA Koichi
Grant type：Competitive
-
性能と生産性を両立するエクサスケールコンピュータ向け階層型粒子法フレームワーク
Grant number：16H02827 2016.4 - 2019.3
日本学術振興会 科学研究費助成事業 基盤研究(B) 基盤研究(B)
丸山 直也, 横田 理央, 田浦 健次朗
Grant amount：\16900000 （ Direct Cost: \13000000 、 Indirect Cost：\3900000 ）
今日のペタスケールシステムの千倍の性能を目指したエクサスケールスーパーコンピュータでは、計算機アーキテクチャの質的および量的な変化が不可避であり、それに従って既存のアプリケーションの大幅な書き換えが必須となる。本研究では頻出基本数値計算手法である粒子法に着目し、アーキテクチャの変更の度にアプリケーションを変更することなく高性能を達成可能なソフトウェア基盤技術を確立することを目的として研究開発を進めた。これは、アーキテクチャ非依存にアプリケーション開発が可能なプログラミングフレームワークに基づき、並列化および性能最適化を自動化することを狙ったものである。今年度は本フレームワークの第一版としてCPUおよびGPUに対応したフレームワークを開発した。本フレームワークはC++テンプレートメタプログラミングに基づき、FMM等の階層的粒子法を簡便に記述可能なプログラミングモデルを提供する。ユーザプログラムはテンプレート展開によってCPU用の並列コードやCUDAを用いたGPU用コードへと自動的に変換されるため、対象プロセッサ用に別途プログラムを作成する必要がない。また、MPIを用いた複数ノード向け並列化もフレームワークによって自動的になされるため、単一のユーザプログラムによって単一プロセッサからスーパーコンピュータクラスの大規模システムまで統一的に動作させることが可能である。また、本フレームワークの実装には高性能を達成するためにFMMアルゴリズムの高性能実装技術に関する研究成果や軽量マルチスレッドランタイムであるMassiveThreadsが活用されており、人手による実装に近い性能が自動的に達成できている。
-
Grant-in-Aid for Scientific Research (B)
2016.4 - 2019.3
Japan Society for the Promotion of Science Grants-in-Aid for Scientific Research
MARUYAMA Naoya
Grant type：Competitive
-
Large scale iterative solvers by combining FMM and H-matrices
Grant number：16H05859 2016.4 - 2018.3
Japan Society for the Promotion of Science Grants-in-Aid for Scientific Research Grant-in-Aid for Young Scientists (A) Grant-in-Aid for Young Scientists (A)
Yokota Rio, Li Xiaoye S., Keyes David E.
Grant amount：\6630000 （ Direct Cost: \5100000 、 Indirect Cost：\1530000 ）
In FY2016, we extended the FMM to H-matrices and developed a LU decomposition code using H-matrices. The dual tree traversal of exaFMM was used to determine the block cluster tree for arbitrary admissibility conditions, which allowed tasked based parallelization of the compression part of the H-matrix code. In FY2017, we further optimized inner kernels of the H-matrix code and compared H-matrices with multigrid for real applications. The use of batched MAGMA enabled us to maximize the performance of GPUs even for small matrices. The advantage of H-matrices over multigrid depends on the condition number of the matrix, while the H-matrix becomes advantageous as the degree of parallelism increases.
-
Grant-in-Aid for Encouragement of Young Scientists (A)
2016.4 - 2018.3
Japan Society for the Promotion of Science Grants-in-Aid for Scientific Research
YOKOTA Rio
Authorship：Principal investigator Grant type：Competitive
-
Grant-in-Aid for Research Activity start-up
2015.8 - 2017.3
Japan Society for the Promotion of Science Grants-in-Aid for Scientific Research
YOKOTA Rio
Authorship：Principal investigator Grant type：Competitive
-
エクサスケーラブルな大規模連立一次方程式の前処理としてのFMMの代数学的拡張
Grant number：15H06196 2015.8 - 2016.3
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 研究活動スタート支援
横田 理央
Grant amount：\1430000 （ Direct Cost: \1100000 、 Indirect Cost：\330000 ）
次世代計算機上で既に性能が出ると分かっている階層的 N 体アルゴリズムを出発点にとり，それを任意の連立一次方程式を扱えるソルバへと徐々に拡張した．平成27年度には，Poisson 方程式しか解くことのできない現在の FMM を Helmholtz 方程式や Stokes 方程式へと拡張し，流体解析のみならず構造・電磁場・音響解析へも適用できるようにした。また、それぞれの方程式を同等の計算条件，計算機環境の下で multigrid 法やHSS行列と直接比較し，いままでやられてこなかった手法間の定量的な優位性の評価を行った．Multigrid 法との比較においては Poisson 方程式に比べ Helmholtz 方程式は FMM の優位性が顕著であった．これは Helmholtz 方程式が高周波を含む場合に multigrid 法の収束性が著しく低下するのに対して，FMM の収束性がさほど低下しないことが原因である．HSS行列との比較ではセットアップのオーバーヘッドが小さい FMM が HSS に比べて合計の計算時間で有利になるという結果が得られた．特に２次元 Laplace 方程式においてその差は顕著で FMM が約1000倍高速であった．スケーラビリティのベンチマークにおいては FMM は Cray XC40 の 131,072 コアを用いた計算で良好な並列化効率が得られ，4000億点規模の計算を数秒で行うことができた．これは，FMM の計算としては世界最大規模であり，最速の計算でもあると思われる．
Teaching Experience
-
高性能科学技術計算
-
計算機ネットワーク
-
5th Academic Group Literacy
Institution：Tokyo Institute of Technology, School of Computing, Department of Computer Science
-
High Performance Scientific Computing
Institution：Tokyo Institute of Technology, School of Computing, Department of Computer Science
-
Computer Networks
Institution：Tokyo Institute of Technology, School of Computing, Department of Computer Science
-
5類リテラシ
Media Coverage
-
“Japan-specific LLMs Are Absolutely Necessary”: A Supercomputer Researcher Pursues Open AI Models Internet
Nikkei xTECH Nikkei xTECH online 2023.7
-
Alarm over Japan Falling Behind in Domestic Generative AI Newspaper, magazine
Nikkei (The Nikkei) Nikkei (The Nikkei) morning edition, page 2 2023.6
-
Growing Participation in Japanese Homegrown Generative AI Newspaper, magazine
Yomiuri Shimbun Yomiuri Shimbun morning edition, page 10 2023.6
-
Tokyo Tech and Others to Develop Domestic Generative AI Newspaper, magazine
Nikkei Sangyo Shimbun Nikkei Sangyo Shimbun page 7 2023.6
-
Pioneering Text-Generating AI with a Domestic Model Newspaper, magazine
Nikkei (The Nikkei) Nikkei (The Nikkei) morning edition, page 14 2023.5
-
Toward the Development of a Japanese Homegrown Conversational AI: Tokyo Tech, RIKEN, and Others Newspaper, magazine
Asahi Shimbun Asahi Shimbun morning edition, page 27 2023.5
-
Japanese Homegrown Generative AI on Fugaku Newspaper, magazine
Nikkei (The Nikkei) Nikkei (The Nikkei) morning edition, page 1 2023.5
-
Developing an LLM on Fugaku Newspaper, magazine
Nikkan Kogyo Shimbun Nikkan Kogyo Shimbun morning edition, page 23 2023.5
-
Toward Developing Domestic Generative AI on the Fugaku Supercomputer Newspaper, magazine
Tokyo Shimbun Tokyo Shimbun morning edition, page 2 2023.5
-
Toward Developing Domestic Generative AI Technology Using Fugaku Newspaper, magazine
Mainichi Shimbun Mainichi Shimbun morning edition, page 18 2023.5
-
Japanese-Language AI Development Begins This Month Newspaper, magazine
Yomiuri Shimbun Yomiuri Shimbun morning edition, page 29 2023.5
-
Domestic Generative AI on Fugaku Newspaper, magazine
Sankei Shimbun Sankei Shimbun morning edition, page 2 2023.5
-
Developing Japanese-Language AI on Fugaku Newspaper, magazine
Yomiuri Shimbun Yomiuri Shimbun evening edition, page 10 2023.5