原 一之
| 生産工学部 電気電子工学科 | 特任教授 |
研究者情報
プロモーション
プロフィール
科学・技術, 情報統計力学、特にオンライン学習で階層型モデルの解析をおこなう。その他、人工知能の基礎技術であるDeep Network の学習特性の解析やDeep neworkを用いた隠れた物体の認識なども研究している。, オンライン学習の統計力学は多くの人工知能の能力の解析に有効である。また、階層型ネットワークのオンライン学習やドロップアウトの理論解析を統計力学の枠組みで行なっている。これらの研究は世界でも数人の研究者しかいない。
学歴
研究活動情報
受賞
- 電気情報通信学会, 2020年LOIS研究賞
マルチエージェントシミュレーションによるCOVID19接触確認アプリ COCOAの感染者数削減効果の検証, 学会誌・学術雑誌による顕彰
大前佑斗、豊谷純、原一之、権寧博、高橋弘毅 - Joint 7th International Conference on Soft Computing and Intelligent Systems and 15th International Symposium on Advanced Intelligent Systems, Best Poster Award
Mutual Leraning for Nonlinear Perceptron, 国際学会・会議・シンポジウム等の賞
Daisuke Saitoh;Kazuyuki Hara
論文
- ★Analysis of Function of Rectified Linear Unit Used in Deep learning
Kazuyuki Hara; Daisuke Saito; and Hayaru Shouno
2015 Internation Joint Conference on Neural Networks, 2015年07月, 査読有り, 通常論文
筆頭著者 - Theoretical Analysis of the SIRVVD Model for Insights Into the Target Rate of COVID-19/SARS-CoV-2 Vaccination in Japan
Yuto Omae; Makoto Sasaki; Jun Toyotani; Kazuyuki Hara; Hirotaka Takahashi
IEEE Access, 2022年 - Impact of removal strategies of stay-at-home,orders on the number of COVID-19 infectors and people leaving their,homes,
Yuto Omae; Yohei Kakimoto; Jun Toyotani; Kazuyuki Hara; Yasuhiro Gon; Hirotaka Takahashi
International Journal of Innovative Computing, Information and,Control, 2021年, 査読有り, 通常論文 - Node-perturbation Learning Applied,for Soft-committee Machine
Kazuyuki Hara; Kentaro Katahira; Masato Okada
IPSJ Transactions on Mathematical Modeling and Its Applications, 2020年08月, 査読有り, 通常論文
筆頭著者 - 染症病床リソースと外出自粛を導入したマルチエージェント環境によるウィルス感染症の伝播予測手法
大前 佑斗; 豊谷 純; 原 一之; 高橋 弘毅
知能と情報, 2020年, 査読有り, 通常論文 - Analysis of Conventional Dropout and its Application to Group Dropout
Kazuyuki Hara; Daisuke Saitoh; Satoshi Suzuki; Takumi Kondou; and Hayaru Shono
IPSJ transaction on Mathematical Modeling and Its Applications, 2017年07月, 査読有り, 通常論文
筆頭著者 - Statistical Mechanics of Node-perturbation Learning with Noisy Baseline
Kazuyuki Hara; Kentaro Katahira; Masato Okada
JOURNAL OF THE PHYSICAL SOCIETY OF JAPAN, 2017年02月, 査読有り, 通常論文 - Group Dropout Inspired by Ensemble Learning
Kazuyuki Hara; Daisuke Saitoh; Takumi Kondou; Satoshi Suzuki; Hayaru Shouno
Lecture Notes in Computer Science, Proc. of ICONIP, 2016年11月, 査読有り, 通常論文 - 22aBT-7 ドロップアウトの正則化効果に関する研究
近藤 佑; 鈴木 聡志; 斎藤 大輔; 原 一之; 庄野 逸
日本物理学会講演概要集, 2016年 - Analysis of Dropout Learning Regarded as Ensemble Learning
Kazuyuki Hara; Daisuke Saitoh; Hayaru Shouno
ARTIFICIAL NEURAL NETWORKS AND MACHINE LEARNING - ICANN 2016, PT II, 2016年, 査読有り, 通常論文 - Mutual Learning using Nonlinear Perceptron
斎藤 大輔、原 一之
Journal of Artificial Intelligence and Soft Computing Research, 2015年04月, 査読有り, 招待有り
ラスト(シニア)オーサー - 22aBL-1 コンボリュージョンネットワークを用いたオンライン学習の収束性に関する研究
斎藤 大輔; 原 一之; 庄野 逸
日本物理学会講演概要集, 2015年 - Improving the convergence property of soft committee machines by replacing derivative with truncated gaussian function
Kazuyuki Hara; Kentaro Katahira
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2014年, 査読有り, 通常論文 - Theoretical analysis of learning speed in gradient descent algorithm replacing derivative with constant
Kazuyuki Hara; Kentaro Katahira
IPSJ Online Transactions, 2014年, 査読有り, 通常論文 - Soft Committee Machine Using Simple Derivative Term
Kazuyuki Hara; Kentaro Katahira
ARTIFICIAL INTELLIGENCE AND SOFT COMPUTING ICAISC 2014, PT I, 2014年, 査読有り, 通常論文 - Mutual Learning Using Nonlinear Perceptron
Daisuke Saitoh; Kazuyuki Hara
2014 JOINT 7TH INTERNATIONAL CONFERENCE ON SOFT COMPUTING AND INTELLIGENT SYSTEMS (SCIS) AND 15TH INTERNATIONAL SYMPOSIUM ON ADVANCED INTELLIGENT SYSTEMS (ISIS), 2014年, 査読有り, 通常論文 - Theoretical analysis of learning speed in gradient descent algorithm replacing derivative with constant
Kazuyuki Hara amd Kentaro Katahira
Information Processing Society of Japan Transactions on Mathematical Modeling and Its Applications, 2013年12月, 査読有り, 通常論文
筆頭著者 - Statistical mechanics of node-perturbation learning for nonlinear perceptron
Kazuyuki Hara; Kentaro Katahira; Kazuo Okanoya; Masato Okada
Journal of the Physical Society of Japan, 2013年05月, 査読有り, 通常論文 - Statistical Mechanics of On-line Ensemble Teacher Learning through a Novel Perceptron Learning Rule
Kazuyuki Hara; Seiji Miyoshi
JOURNAL OF THE PHYSICAL SOCIETY OF JAPAN, 2012年06月, 査読有り, 通常論文 - Theoretical analysis of function of derivative term in on-line gradient descent learning
Kazuyuki Hara; Kentaro Katahira; Kazuo Okanoya; Masato Okada
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2012年, 査読有り, 通常論文 - Ensemble-Teacher Learning through a Perceptron Rule with a Margin
Kazuyuki Hara; Seiji Miyoshi
ARTIFICIAL NEURAL NETWORKS AND MACHINE LEARNING - ICANN 2011, PT I, 2011年, 査読有り, 通常論文 - Statistical Mechanics of On-line Node-perturbation Learning
Kentaro Katahira; Kazuo Okanoya; Masato Okada
Information Processing Society of Japan Transactions on Mathematical Modeling and Its Applications, 2011年01月, 査読有り, 通常論文
筆頭著者 - On-Line Ensemble-Teacher Learning through a Perceptron Rule with a Margin
Kazuyuki Hara; Katsuya Ono; Seiji Miyoshi
ARTIFICIAL NEURAL NETWORKS (ICANN 2010), PT III, 2010年, 査読有り, 通常論文 - Statistical Mechanics of On-Line Mutual Learning with Many Linear Perceptrons
Kazuyuki Hara; Yoichi Nakayama; Seiji Miyoshi; Masato Okada
JOURNAL OF THE PHYSICAL SOCIETY OF JAPAN, 2009年11月, 査読有り, 通常論文 - Analysis of Ising Spin Neural Network with Time-Dependent Mexican-Hat-Type Interaction
Kazuyuki Hara; Seiji Miyoshi; Tatsuya Uezu; Masato Okada
ADVANCES IN NEURO-INFORMATION PROCESSING, PT II, 2009年, 査読有り, 通常論文 - Mutual Learning with Many Linear Perceptrons: On-Line Learning Theory
Kazuyuki Hara; Yoichi Nakayama; Seiji Miyoshi; Masato Okada
ARTIFICIAL NEURAL NETWORKS - ICANN 2009, PT I, 2009年, 査読有り, 通常論文 - Optimization of the asymptotic property of mutual learning involving an integration mechanism of ensemble learning
Kazuyuki Hara; Takahiro Yamada
JOURNAL OF THE PHYSICAL SOCIETY OF JAPAN, 2008年02月, 査読有り, 通常論文 - Statistical mechanics of mutual learning with a latent teacher
Kazuyuki Hara; Masato Okada
JOURNAL OF THE PHYSICAL SOCIETY OF JAPAN, 2007年01月, 査読有り, 通常論文 - Analysis of ensemble learning using simple perceptrons based on on-line learning theory
Seiji Miyoshi; Kazuyuki Hara; Masato Okada
Systems and Computers in Japan, 2005年11月, 査読有り, 通常論文 - Ensemble Learning of Linear Perceptrons: On-Line Learning Theory
Masato Okada
Journal of the Physical Society of Japan, 2005年11月, 査読有り, 通常論文
筆頭著者 - Analysis of ensemble learning using simple perceptrons based on online learning theory
S Miyoshi; K Hara; M Okada
PHYSICAL REVIEW E, 2005年03月, 査読有り, 通常論文 - Analysis of ensemble learning using simple perceptrons based on online learning theory
S Miyoshi; K Hara; M Okada
PROGRESS OF THEORETICAL PHYSICS SUPPLEMENT, 2005年, 査読有り, 通常論文 - On-line leaning of a simple perceptron learning with margin.
Kazuyuki Hara; Masato Okada
Margin Systems and Computers in Japan, 2004年06月, 査読有り, 通常論文 - On-line learning through simple perceptron learning with a margin
K Hara; M Okada
NEURAL NETWORKS, 2004年03月, 査読有り, 通常論文 - Online learning theory of ensemble learning using linear perceptrons
Kazuyuki Hara; Masato Okada
IEEE International Conference on Neural Networks - Conference Proceedings, 2004年, 査読有り, 通常論文 - On-line learning through simple perceptron learning with a margin
K Hara; M Okada
ICONIP'02: PROCEEDINGS OF THE 9TH INTERNATIONAL CONFERENCE ON NEURAL INFORMATION PROCESSING, 2002年, 査読有り, 通常論文 - 階層形神経回路網と線形信号処理法の信号分類能力の比較
原一之; 中山謙二
情報処理学会論文誌, 1997年, 査読有り, 通常論文
筆頭著者
MISC
- パーシャルアニーリングの統計力学 : 相互作用がメキシカンハット型の場合(ハードウェア(2),ニューロハードウェア,一般)
原 一之; 上江洌 達也; 三好 誠司; 岡田 真人
電子情報通信学会技術研究報告. NC, ニューロコンピューティング, 2008年10月31日, 査読無し, 通常論文 - 22pVC-6 メキシカンハット型相互作用が時間変化する系のレプリカ解析II(22pVC 情報統計力学,領域11(統計力学,物性基礎論,応用数学,力学,流体物理))
原 一之; 上江洌 達也; 三好 誠司; 岡田 真人
日本物理学会講演概要集, 2008年08月25日, 査読無し, 通常論文 - 21pWB-1 擬似教師つき学習とアンサンブル学習(ニューラルネットワーク,領域11,統計力学,物性基礎論,応用数学,力学,流体物理)
岡田 真人; 原 一之; 三好 誠司
日本物理学会講演概要集, 2007年02月28日, 査読無し, 通常論文 - 21aWA-5 エージェント数が任意な場合の相互学習(情報統計力学,領域11,統計力学,物性基礎論,応用数学,力学,流体物理)
原 一之; 岡田 真人
日本物理学会講演概要集, 2007年02月28日, 査読無し, 通常論文 - アンサンブル学習の統合機構と相互学習
原 一之; 岡田 真人
電子情報通信学会技術研究報告. NC, ニューロコンピューティング, 2006年03月09日, 査読無し, 通常論文 - 29pXH-14 潜在的な教師のある相互学習の統計力学(29pXH ニューラルネットワーク(神経系のモデルを含む),領域11(統計力学,物性基礎論,応用数学,力学,流体物理))
原 一之; 岡田 真人
日本物理学会講演概要集, 2006年03月04日, 査読無し, 通常論文 - 潜在的な教師のある相互学習の統計力学
原 一之; 岡田 真人
電子情報通信学会技術研究報告. NC, ニューロコンピューティング, 2005年06月16日, 査読無し, 通常論文 - 教師が非単調な場合のアンサンブル学習
三好 誠司; 原 一之; 岡田 真人
電子情報通信学会技術研究報告. NC, ニューロコンピューティング, 2005年03月23日, 査読無し, 通常論文 - 24aYB-6 教師が非単調な場合のアンサンブル学習(情報統計力学・ニューラルネットワーク,領域11(統計力学,物性基礎論,応用数学,力学,流体物理))
三好 誠司; 原 一之; 岡田 真人
日本物理学会講演概要集, 2005年03月04日, 査読無し, 通常論文 - 線形ウィークラーナーによるアンサンブル学習の汎化誤差の解析
原 一之; 岡田 真人
システム制御情報学会論文誌, 2004年12月15日, 査読無し, 通常論文 - 14pTD-4 教師がコミティマシンの場合のアンサンブル学習(情報統計力学, 領域 11)
三好 誠司; 原 一之; 岡田 真人
日本物理学会講演概要集, 2004年08月25日, 査読無し, 通常論文 - 14pTD-5 学習過程のノイズがパラレルブースティングに与える影響(情報統計力学, 領域 11)
原 一之; 岡田 真人
日本物理学会講演概要集, 2004年08月25日, 査読無し, 通常論文 - オンライン学習理論に基づく単純パーセプトロンのアンサンブル学習の解析(パターン認識)
三好 誠司; 原 一之; 岡田 真人
電子情報通信学会論文誌. D-II, 情報・システム, II-パターン処理, 2004年07月01日, 査読無し, 通常論文 - 21pTQ-1 オンライン学習理論に基づく非線形単純パーセプトロンのアンサンブル学習の解析
三好 誠司; 原 一之; 岡田 真人
日本物理学会講演概要集, 2003年08月15日, 査読無し, 通常論文 - [チュートリアル講演]アンサンブル学習(<特集>統計的学習理論及び一般)
岡田 真人; 原 一之; 三好 誠司
電子情報通信学会技術研究報告. NC, ニューロコンピューティング, 2003年07月22日, 査読無し, 通常論文 - オンライン学習理論に基づく単純パーセプトロンのアンサンブル学習の解析(<特集>統計的学習理論及び一般)
三好 誠司; 原 一之; 岡田 真人
電子情報通信学会技術研究報告. NC, ニューロコンピューティング, 2003年07月22日, 査読無し, 通常論文 - パラレルブースティングのオンラインラーニングの理論
原 一之; 岡田 真人
電子情報通信学会技術研究報告. NC, ニューロコンピューティング, 2003年06月20日, 査読無し, 通常論文 - 28aWJ-10 パラレルブースティングのオンラインラーニングの理論
原 一之; 岡田 真人
日本物理学会講演概要集, 2003年03月06日, 査読無し, 通常論文 - 学習の問題を統計力学で取り扱う:線形パーセプトロンのアンサンブル学習を一例として (特集 情報論的学習理論--機械学習のさまざまな形)
岡田 真人; 原 一之
Computer today, 2003年03月, 査読無し, 通常論文 - マージンを用いた単純パーセプトロン学習法のオンラインラーニングの理論
原 一之; 岡田 真人
電子情報通信学会論文誌. D-II, 情報・システム, II-パターン処理, 2002年10月01日, 査読有り, 通常論文
筆頭著者 - マージンを用いた単純パーセプトロン学習法のオンラインラーニングの理論
原 一之; 岡田 真人
電子情報通信学会技術研究報告. NC, ニューロコンピューティング, 2001年12月14日, 査読無し, 通常論文
講演・口頭発表等
- Performance of pre-learned convolution neural networks applied to recognition of overlapping digits
Dig SHII; Ryosuke MIYOSHI; and Kazuyuki Hara
2020 IEEE International Conference on Big Data and Smart Computing (BigComp), 2020年02月, Korean Institute of Information Scientists and Engineers - Empirical Study of Effect of Dropout in Online Learning
Kazuyuki Hara
26th International Conference on Artificial Neural Networks, 2017年09月, Europian Neural Network Society, 通常論文 - Analysis of Dropout Learning Regarded as Ensemble Learning
Kazuyuki Hara; Daisuke SAITOH; and Hayaru SHOUNO
25th international conference on artificial neural networks, 2016年09月, European Neural Network Society, 通常論文 - ソフトコミッティマシンのノードパータベーション学習
原 一之
電子情報通信学会ニューラルコンピューティング研究会, 2016年01月, 電子情報通信学会ニューラルコンピューティング研究会, 通常論文 - 新しいDropout 法の提案と動特性の解析
斎藤 大輔
電子情報通信学会 ニューラルコンピューティング研究会, 2016年01月, 電子情報通信学会 ニューラルコンピューティング研究会, 通常論文 - Dropout as an ensemble learning
Kazuyuki Hara
International Meeting on "High-dimensional Data Driven Science, 2015年12月, 文部科学省科学研究費補助金「新学術領域研究」スパースモデリングの深化と高次元データ駆動科学の創成, 通常論文 - Dropoutは対称性の破壊を加速するのか?
斎藤 大輔、原 一之
第25回 日本神経回路学会 全国大会, 2015年09月, 日本神経回路学会, 通常論文 - Model Compressionの実験的評価-真のモデルが既知の場合-
近藤 佑、日野 英逸、原 一之
第25回 日本神経回路学会 全国大会, 2015年09月, 日本神経回路学会, 通常論文 - Analysis of Function of Rectified Linear Unit Used in Deep learning
Kazuyuki Hara; Daisuke Saitoh; and Hayaru Shouno
The International Joint Conference on Neural Networks, 2015年07月, International Neural Network Society, IEEE Computational Intelligence Society, 通常論文 - Mutual Learning Using Nonlinear Perceptron
Daisuke Saitoh; Kazuyuki Hara
Joint 7th International Conference on Soft Computing and Intelligent Systems and 15th International Symposium on Advanced Intelligent Systems, 2014年12月, Japan Society for Fuzzy Theory and Intelligence Informatics, 通常論文 - Improving the Convergence Property of Soft Committee Machines by Replacing Derivative with Truncated Gaussian Function
Kazuyuki Hara
The 24th International Conference on Artificial Neural Networks, 2014年09月, The European Neural Network Society, 通常論文 - Improving the Convergence Property of SoftCommittee Machines by Replacing Derivative with Truncated Gaussian Function
Kazuyuki Hara
The 24th International Conference on Artificial Neural Networks, 2014年09月, The European Neural Network Society, 通常論文 - Soft Committee Machine Using Simple Derivative Term
Kazuyuki Hara
The 13th International Conference on Artificial Intelligence and Soft computing, 2014年06月, Polish Neural Network Society, 通常論文
共同・受託研究希望テーマ
共同研究・競争的資金等の研究課題
- 深層学習によるCOVID-19感染伝搬と経済活動を同時制御する社会運営戦略の発見
日本学術振興会, 科学研究費補助金 基盤研究(C), 2021年04月 - 2025年03月
豊谷純、大前佑斗、原一之、高橋弘毅 - オンライン学習におけるドロップアウトの理論と最適化に関する研究
独立行政法人日本学術振興会, 科学研究費補助金 基盤(C), 2018年04月 - 2023年03月
原 一之 - 多数の学習機械の協調メカニズムとその動的過程に関する研究
日本学術振興会, 科学研究費助成事業, 2004年 - 2006年
原 一之; 三好 誠司 - 階層形ニューラルネットワーク学習の収束性の改善とサーチエンジンへの応用
日本学術振興会, 科学研究費助成事業, 2001年 - 2002年
原 一之; 中山 謙二