|
Affiliation Department |
工学専攻 情報工学系プログラム知能情報分野 |
|
Homepage |
|
|
External Link |
|
TANAKA Gouhei
|
|
|
Research Interests
-
Dynamical systems
-
Control
-
Bifurcation theory
-
Artificial intelligence
-
Reservoir computing
Research Areas
-
Informatics / Soft computing
-
Informatics / Mathematical informatics
From School
-
The University of Tokyo Graduate School of Frontier Sciences Department of Complexity Science and Engineering, Doctoral course Graduated
2002.04 - 2005.03
-
The University of Tokyo Graduate School of Frontier Sciences Department of Complexity Science and Engineering, Master course Graduated
2000.04 - 2002.03
-
The University of Tokyo Faculty of Engineering Department of Mathematical Engineering and Information Physics Graduated
1998.04 - 2000.03
-
The University of Tokyo Faculty of Liberal Arts Natural Sciences 1 Graduated
1996.04 - 1998.03
From Graduate School
-
The University of Tokyo Graduate School of Frontier Sciences Department of Complexity Science and Engineering Doctor's Course Completed
- 2005.03
-
The University of Tokyo Graduate School of Frontier Sciences Department of Complexity Science and Engineering Master's Course Completed
- 2002.03
External Career
-
Nagoya Institute of Technology Professor
2023.04
-
MEXT Superrobust Computation Project, Information Science and Technology Strategic Core Research Assistant
2002.12 - 2005.03
-
The University of Tokyo Institute of Industrial Science Assistant Professor
2007.04 - 2011.07
-
The University of Tokyo Institute of Industrial Science Project Associate Professor
2011.08 - 2013.03
-
The University of Tokyo Institute of Industrial Science Assistant Professor
2006.08 - 2007.03
Papers
-
Long-time-constant leaky-integrating oxygen-vacancy drift-diffusion FET for human-interactive spiking reservoir computing. Reviewed
Hisashi Inoue, Hiroto Tamura, Ai Kitoh, Xiangyu Chen, Zolboo Byambadorj, Takeaki Yajima, Yasushi Hotta, Tetsuya Iizuka, Gouhei Tanaka, Isao H. Inoue
2023 IEEE Symposium on VLSI Technology and Circuits 1 - 2 2023.07
Language：English Publishing type：Research paper (international conference proceedings)
DOI： 10.23919/VLSITechnologyandCir57934.2023.10185412
Other Link： https://ieeexplore.ieee.org/document/10185412
-
Performance Enhancement of a Spin-Wave-Based Reservoir Computing System Utilizing Different Physical Conditions Reviewed
Ryosho Nakane, Akira Hirose, Gouhei Tanaka
Physical Review Applied 19 ( 3 ) 2023.03
Authorship：Last author Language：English Publishing type：Research paper (scientific journal)
We numerically study how to enhance reservoir computing performance by thoroughly extracting the spin-wave device potential for higher-dimensional information generation. The reservoir device has a 1-input exciter and 120-output detectors on the top of a continuous magnetic garnet film for spin-wave transmission. For various nonlinear and fading-memory dynamic phenomena distributing in the film space, small in-plane magnetic fields are used to prepare stripe domain structures and various damping constants at the film sides and bottom are explored. The ferromagnetic resonant frequency and relaxation time of spin precession clearly characterizes the change in spin dynamics with the magnetic field and damping constant. The common input signal for reservoir computing is a 1-GHz cosine wave with random 6-valued amplitude modulation. A basic 120-dimensional reservoir output vector is obtained from time-series signals at the 120-output detectors under each of three magnetic field conditions. Then, 240- and 360-dimensional reservoir output vectors are also constructed by concatenating two and three basic ones, respectively. In nonlinear autoregressive moving average (NARMA) prediction tasks, the computational performance is enhanced as the dimension of the reservoir output vector becomes higher and a significantly low prediction error is achieved for the tenth-order NARMA task using the 360-dimensional vector and optimum damping constant. The results are clear evidence that the collection of diverse output signals efficiently increases the dimensionality of the integrated reservoir state vector (i.e. reservoir-state richness) and thereby contributes to high computational performance. This paper demonstrates that performance enhancement through various configuration settings is a practical approach for on-chip reservoir computing devices with small numbers of real output nodes.
-
Multi-Reservoir Echo State Networks with Hodrick-Prescott Filter for nonlinear time-series prediction.
Ziqiang Li, Yun Liu, Gouhei Tanaka
Applied Soft Computing 135 110021 - 110021 2023.03
Authorship：Last author, Corresponding author Language：English Publishing type：Research paper (scientific journal)
-
Spatially Arranged Sparse Recurrent Neural Networks for Energy Efficient Associative Memory. Reviewed International journal
Gouhei Tanaka, Ryosho Nakane, Tomoya Takeuchi, Toshiyuki Yamane, Daiju Nakano, Yasunao Katayama, Akira Hirose
IEEE transactions on neural networks and learning systems 31 ( 1 ) 24 - 38 2020.01
Authorship：Lead author, Corresponding author Language：English Publishing type：Research paper (scientific journal)
The development of hardware neural networks, including neuromorphic hardware, has been accelerated over the past few years. However, it is challenging to operate very large-scale neural networks with low-power hardware devices, partly due to signal transmissions through a massive number of interconnections. Our aim is to deal with the issue of communication cost from an algorithmic viewpoint and study learning algorithms for energy-efficient information processing. Here, we consider two approaches to finding spatially arranged sparse recurrent neural networks with the high cost-performance ratio for associative memory. In the first approach following classical methods, we focus on sparse modular network structures inspired by biological brain networks and examine their storage capacity under an iterative learning rule. We show that incorporating long-range intermodule connections into purely modular networks can enhance the cost-performance ratio. In the second approach, we formulate for the first time an optimization problem where the network sparsity is maximized under the constraints imposed by a pattern embedding condition. We show that there is a tradeoff between the interconnection cost and the computational performance in the optimized networks. We demonstrate that the optimized networks can achieve a better cost-performance ratio compared with those considered in the first approach. We show the effectiveness of the optimization approach mainly using binary patterns and apply it also to gray-scale image restoration. Our results suggest that the presented approaches are useful in seeking more sparse and less costly connectivity of neural networks for the enhancement of energy efficiency in hardware neural networks.
DOI： 10.1109/TNNLS.2019.2899344
Other Link： https://dblp.uni-trier.de/db/journals/tnn/tnn31.html#TanakaNTYNKH20
-
Recent advances in physical reservoir computing: A review Reviewed International journal
Gouhei Tanaka, Toshiyuki Yamane, Jean Benoit Heroux, Ryosho Nakane, Naoki Kanazawa, Seiji Takeda, Hidetoshi Numata, Daiju Nakano, Akira Hirose
NEURAL NETWORKS 115 100 - 123 2019.07
Authorship：Lead author, Corresponding author Language：English Publishing type：Research paper (scientific journal) Publisher：PERGAMON-ELSEVIER SCIENCE LTD
Reservoir computing is a computational framework suited for temporal/sequential data processing. It is derived from several recurrent neural network models, including echo state networks and liquid state machines. A reservoir computing system consists of a reservoir for mapping inputs into a high-dimensional space and a readout for pattern analysis from the high-dimensional states in the reservoir. The reservoir is fixed and only the readout is trained with a simple method such as linear regression and classification. Thus, the major advantage of reservoir computing compared to other recurrent neural networks is fast learning, resulting in low training cost. Another advantage is that the reservoir without adaptive updating is amenable to hardware implementation using a variety of physical systems, substrates, and devices. In fact, such physical reservoir computing has attracted increasing attention in diverse fields of research. The purpose of this review is to provide an overview of recent advances in physical reservoir computing by classifying them according to the type of the reservoir. We discuss the current issues and perspectives related to physical reservoir computing, in order to further expand its practical applications and develop next-generation machine learning systems.
DOI： 10.1016/j.neunet.2019.03.005
Other Link： https://dblp.uni-trier.de/db/journals/nn/nn115.html#TanakaYHNKTNNH19
-
Random and targeted interventions for epidemic control in metapopulation models. Reviewed International journal
Gouhei Tanaka, Chiyori Urabe, Kazuyuki Aihara
Scientific reports 4 5522 - 5522 2014.07
Authorship：Lead author, Corresponding author Language：English Publishing type：Research paper (scientific journal) Publisher：NATURE PUBLISHING GROUP
In general, different countries and communities respond to epidemics in accordance with their own control plans and protocols. However, owing to global human migration and mobility, strategic planning for epidemic control measures through the collaboration of relevant public health administrations is gaining importance for mitigating and containing large-scale epidemics. Here, we present a framework to evaluate the effectiveness of random (non-strategic) and targeted (strategic) epidemic interventions for spatially separated patches in metapopulation models. For a random intervention, we analytically derive the critical fraction of patches that receive epidemic interventions, above which epidemics are successfully contained. The analysis shows that the heterogeneity of patch connectivity makes it difficult to contain epidemics under the random intervention. We demonstrate that, particularly in such heterogeneously connected networks, targeted interventions are considerably effective compared to the random intervention. Our framework is useful for identifying the target areas where epidemic control measures should be focused.
DOI： 10.1038/srep05522
-
Dynamical robustness of coupled heterogeneous oscillators. Reviewed International journal
Gouhei Tanaka, Kai Morino, Hiroaki Daido, Kazuyuki Aihara
Physical review. E, Statistical, nonlinear, and soft matter physics 89 ( 5 ) 052906 - 052906 2014.05
Authorship：Lead author, Corresponding author Language：English Publishing type：Research paper (scientific journal) Publisher：AMER PHYSICAL SOC
We study tolerance of dynamic behavior in networks of coupled heterogeneous oscillators to deterioration of the individual oscillator components. As the deterioration proceeds with reduction in dynamic behavior of the oscillators, an order parameter evaluating the level of global oscillation decreases and then vanishes at a certain critical point. We present a method to analytically derive a general formula for this critical point and an approximate formula for the order parameter in the vicinity of the critical point in networks of coupled Stuart-Landau oscillators. Using the critical point as a measure for dynamical robustness of oscillator networks, we show that the more heterogeneous the oscillator components are, the more robust the oscillatory behavior of the network is to the component deterioration. This property is confirmed also in networks of Morris-Lecar neuron models coupled through electrical synapses. Our approach could provide a useful framework for theoretically understanding the role of population heterogeneity in robustness of biological networks.
-
Dynamical robustness in complex networks: the crucial role of low-degree nodes. Reviewed International journal
Gouhei Tanaka, Kai Morino, Kazuyuki Aihara
Scientific reports 2 232 - 232 2012
Authorship：Lead author, Corresponding author Language：English Publishing type：Research paper (scientific journal) Publisher：NATURE PUBLISHING GROUP
Many social, biological, and technological networks consist of a small number of highly connected components (hubs) and a very large number of loosely connected components (low-degree nodes). It has been commonly recognized that such heterogeneously connected networks are extremely vulnerable to the failure of hubs in terms of structural robustness of complex networks. However, little is known about dynamical robustness, which refers to the ability of a network to maintain its dynamical activity against local perturbations. Here we demonstrate that, in contrast to the structural fragility, the nonlinear dynamics of heterogeneously connected networks can be highly vulnerable to the failure of low-degree nodes. The crucial role of low-degree nodes results from dynamical processes where normal (active) units compensate for the failure of neighboring (inactive) units at the expense of a reduction in their own activity. Our finding highlights the significant difference between structural and dynamical robustness in complex networks.
DOI： 10.1038/srep00232
-
Complex-valued multistate associative memory with nonlinear multilevel functions for gray-level image reconstruction. Reviewed International journal
Gouhei Tanaka, Kazuyuki Aihara
IEEE transactions on neural networks 20 ( 9 ) 1463 - 73 2009.09
Authorship：Lead author, Corresponding author Language：English Publishing type：Research paper (scientific journal)
A widely used complex-valued activation function for complex-valued multistate Hopfield networks is revealed to be essentially based on a multilevel step function. By replacing the multilevel step function with other multilevel characteristics, we present two alternative complex-valued activation functions. One is based on a multilevel sigmoid function, while the other on a characteristic of a multistate bifurcating neuron. Numerical experiments show that both modifications to the complex-valued activation function bring about improvements in network performance for a multistate associative memory. The advantage of the proposed networks over the complex-valued Hopfield networks with the multilevel step function is more outstanding when a complex-valued neuron represents a larger number of multivalued states. Further, the performance of the proposed networks in reconstructing noisy 256 gray-level images is demonstrated in comparison with other recent associative memories to clarify their advantages and disadvantages.
-
Bifurcation analysis on a hybrid systems model of intermittent hormonal therapy for prostate cancer Reviewed
Gouhei Tanaka, Kunichika Tsumoto, Shigeki Tsuji, Kazuyuki Aihara
PHYSICA D-NONLINEAR PHENOMENA 237 ( 20 ) 2616 - 2627 2008.10
Authorship：Lead author, Corresponding author Language：English Publishing type：Research paper (scientific journal) Publisher：ELSEVIER SCIENCE BV
Hybrid systems are widely used to model dynamical phenomena that are characterized by interplay between continuous dynamics and discrete events. An example of biomedical application is modeling of disease progression of prostate cancer under intermittent hormonal therapy, where continuous tumor dynamics is switched by interruption and reinstitution of medication. In the present paper, we study a hybrid systems model representing intermittent androgen suppression (IAS) therapy for advanced prostate cancer. Intermittent medication with switching between on-treatment and off-treatment periods is intended to possibly prevent a prostatic tumor from developing into a hormone-refractory state and is anticipated as a possible strategy for delaying or hopefully averting a cancer relapse which most patients undergo as a result of long-term hormonal suppression. Clinical efficacy of IAS therapy for prostate cancer is still under investigation but at least worth considering in terms of reduction of side effects and economic costs during off-treatment periods. In the model of IAS therapy, it depends on some clinically controllable parameters whether a relapse of prostate cancer occurs or not. Therefore, we examine nonlinear dynamics and bifurcation structure of the model by exploiting a numerical method to clarify bifurcation sets in the hybrid system. Our results suggest that adjustment of the normal androgen level in combination with appropriate medication scheduling could enhance the possibility of relapse prevention. Moreover, a two-dimensional piecewise-linear system reduced from the original model highlights the origin of nonlinear phenomena specific to the hybrid system. (C) 2008 Elsevier B.V. All rights reserved.
Books and Other Publications
-
リザバーコンピューティング : 時系列パターン認識のための高速機械学習の理論とハードウェア
田中, 剛平, 中根, 了昌, 廣瀬, 明（ Role： Joint author）
森北出版 2021.03 （ ISBN:9784627855311 ）
Total pages：vi, 207p Language：jpn Book type：Scholarly book
Misc
-
リザバーコンピューティング Invited
田中剛平
106 ( 6 ) 2023.06
Authorship：Lead author, Last author, Corresponding author Language：Japanese Publishing type：Article, review, commentary, editorial, etc. (scientific journal)
-
リザバーコンピューティングの概念と最近の動向
タナカ ゴウヘイ
電子情報通信学会誌 102 ( 2 ) 108 - 113 2019.02
-
Reservoir computing utilizing spin waves: enhancement of computational performance through a practical approach for on-chip devices
応用物理学会春季学術講演会講演予稿集(CD-ROM) 70th 2023.03
-
Guest Editorial Special Issue on New Frontiers in Extremely Efficient Reservoir Computing
Gouhei Tanaka, Claudio Gallicchio, Alessio Micheli, Juan Pablo Ortega, Akira Hirose
IEEE Transactions on Neural Networks and Learning Systems 33 ( 6 ) 2571 - 2574 2022.06
Publishing type：Book review, literature introduction, etc.
-
Reservoir Computing utilizing spin waves propagating through magnetic domains
中根了昌, 廣瀬明, 田中剛平, 田中剛平, 田中剛平
応用物理学会春季学術講演会講演予稿集(CD-ROM) 69th 2022
-
Reservoir computing with spiking neural networks and reward-modulated STDP
鶴海杭之, 田中剛平, 田中剛平
電子情報通信学会技術研究報告(Web) 122 ( 65(NLP2022 1-25) ) 2022
-
Experiments of Reconstructive Reservoir Computing to Detect Anomaly in Time-series Signals
加藤准也, 田中剛平, 田中剛平, 中根了昌, 廣瀬明
電子情報通信学会技術研究報告(Web) 121 ( 390(NC2021 46-78) ) 2022
-
Relationship between Computational Performance and Task Difficulty of Reinforcement Learning Methods Using Reward Machines
渡邊隆二, 田中剛平
電子情報通信学会技術研究報告(Web) 121 ( 444(NLP2021 126-152) ) 2022
-
Effects of sparse connections in spiking neural networks for unsupervised pattern recognition
品川大樹, 藤原寛太郎, 田中剛平, 田中剛平
電子情報通信学会技術研究報告(Web) 121 ( 444(NLP2021 126-152) ) 2022
Presentations
-
Diverse-Timescale Echo State Networks for Multiscale Modeling International conference
Gouhei Tanaka
International Conference on Neuromorphic, Natural, and Physical Computing 2023.10
Event date： 2023.10
Language：English Presentation type：Oral presentation (general)
Venue：Hannover Country：Germany
-
Tailoring oxygen diffusion dynamics in three-terminal devices for spiking reservoir computing International conference
Hisashi Inoue, H. Tamura, A. Kitoh, X. Chen, Z. Byambadorj, T. Yajima, Y. Hotta, T. Iizuka, G. Tanaka, I. H. Inoue
MemriSys 2023 2023.11
-
リザバーコンピューティングによる時系列データ学習 Invited
田中 剛平
第15回Nagoyaオープンイノベーション研究会 2023.09
Event date： 2023.09
Language：Japanese Presentation type：Oral presentation (invited, special)
-
Mahalanobis Distance of Reservoir States for Highly-Efficient Time-Series Classification
Hiroto Tamura, Kantaro Fujiwara, Kazuyuki Aihara, Gouhei Tanaka
The 33rd Annual Meeting of the Japanese Neural Network Society 2023.09
-
Identity Recognition with Bidirectional Echo State Networks from Continuous Blood Pressure Data
Ziqiang Li, Kantaro Fujiwara, Gouhei Tanaka
The 33rd Annual Meeting of the Japanese Neural Network Society 2023.09
-
リザバーコンピューティングの高性能化・高効率化に関して Invited
田中 剛平
第15回 AI Optics研究グループ研究会 2023.03
Event date： 2023.03
Language：Japanese Presentation type：Oral presentation (invited, special)
-
AIによる複雑ダイナミクスのパターン認識 Invited
田中 剛平
日本生理学会第100回記念大会, 100周年記念事業委員会企画シンポジウム「AIが切り開く医学・生理学・生命科学の新展開」 2023.03
Event date： 2023.03
Language：Japanese Presentation type：Oral presentation (invited, special)
-
リザバーコンピューティングとマルチスケールモデリング Invited
田中 剛平
電子情報通信学会総合大会, チュートリアルセッション「NBT-1. 次世代ネットワークを支える数理モデルの展開」 2023.03
Event date： 2023.03
Language：Japanese Presentation type：Oral presentation (invited, special)
-
Indoor air quality prediction using multi-reservoir echo state network with attention mechanism
Wenrui Qiu, Gouhei Tanaka
2023.01
-
Inference of network structure in dynamic systems using a generative model International conference
Shuhan Zheng, Gouhei Tanaka
Dynamics Days US 2023 2023.01
Awards
-
Best Paper Award 3rd Prize, 29th International Conference on Artificial Neural Networks (ICANN)
2020.11 European Neural Network Society Two-Step FORCE Learning Algorithm for Fast Convergence in Reservoir Computing
Hiroto Tamura, Gouhei Tanaka
-
Best Paper Award
2017 IEEE Transcations on Nanotechnology 2016
Gouhei Tanaka, Yasunao Katayama, Daiju Nakano, Toshiyuki Yamane, Ryosho Nakane
-
Best Poster Award (Silver Prize)
2011 First International Symposium on Innovative Mathematical Modelling Robustness of Scale-free Networks of Oscillators against Partial Inactivation
Gouhei Tanaka
-
Honorable Mention Award, DSWeb Tutorials Contest
2005 Society for Industrial and Applied Mathematics Crisis-induced Intermittency in Coupled Chaotic Maps
Gouhei Tanaka
Committee Memberships
-
Information-technology Promotion Agency, MITOU program Project Manager
2023.08
Committee type：Other
-
日本学術振興会 先導的研究開発委員会「マテリアル・インフォマティクスによるものづくりプラットフォームの戦略的構築」
2016 - 2018
Committee type：Academic society
-
科学技術振興機構 研究開発戦略センター(CRDS) 「研究開発の俯瞰報告書」（システム科学技術分野）
2012 - 2015
Committee type：Academic society