論文 - 田中 剛平
-
Long-time-constant leaky-integrating oxygen-vacancy drift-diffusion FET for human-interactive spiking reservoir computing. 査読あり
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年07月
記述言語：英語 掲載種別：研究論文（国際会議プロシーディングス）
-
Performance Enhancement of a Spin-Wave-Based Reservoir Computing System Utilizing Different Physical Conditions 査読あり
Ryosho Nakane, Akira Hirose, Gouhei Tanaka
Physical Review Applied 19 ( 3 ) 2023年03月
担当区分：最終著者 記述言語：英語 掲載種別：研究論文（学術雑誌）
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年03月
-
Spatially Arranged Sparse Recurrent Neural Networks for Energy Efficient Associative Memory. 査読あり 国際誌
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年01月
担当区分：筆頭著者, 責任著者 記述言語：英語 掲載種別：研究論文（学術雑誌）
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
その他リンク： https://dblp.uni-trier.de/db/journals/tnn/tnn31.html#TanakaNTYNKH20
-
Recent advances in physical reservoir computing: A review 査読あり 国際誌
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年07月
担当区分：筆頭著者, 責任著者 記述言語：英語 掲載種別：研究論文（学術雑誌） 出版者・発行元：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
その他リンク： https://dblp.uni-trier.de/db/journals/nn/nn115.html#TanakaYHNKTNNH19
-
Random and targeted interventions for epidemic control in metapopulation models. 査読あり 国際誌
Gouhei Tanaka, Chiyori Urabe, Kazuyuki Aihara
Scientific reports 4 5522 - 5522 2014年07月
担当区分：筆頭著者, 責任著者 記述言語：英語 掲載種別：研究論文（学術雑誌） 出版者・発行元：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. 査読あり 国際誌
Gouhei Tanaka, Kai Morino, Hiroaki Daido, Kazuyuki Aihara
Physical review. E, Statistical, nonlinear, and soft matter physics 89 ( 5 ) 052906 - 052906 2014年05月
担当区分：筆頭著者, 責任著者 記述言語：英語 掲載種別：研究論文（学術雑誌） 出版者・発行元：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. 査読あり 国際誌
Gouhei Tanaka, Kai Morino, Kazuyuki Aihara
Scientific reports 2 232 - 232 2012年
担当区分：筆頭著者, 責任著者 記述言語：英語 掲載種別：研究論文（学術雑誌） 出版者・発行元：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 Associate Memory with Nonlinear Multilevel Functions for Gray-Level Image Reconstruction 査読あり 国際誌
Gouhei Tanaka, Kazuyuki Aihara
IEEE Transactions on Neural Networks 20 20 ( 9 ) 1463 - 73 2009年09月
-
Bifurcation analysis on a hybrid systems model of intermittent hormonal therapy for prostate cancer 査読あり
Gouhei Tanaka, Kunichika Tsumoto, Shigeki Tsuji, Kazuyuki Aihara
PHYSICA D-NONLINEAR PHENOMENA 237 ( 20 ) 2616 - 2627 2008年10月
担当区分：筆頭著者, 責任著者 記述言語：英語 掲載種別：研究論文（学術雑誌） 出版者・発行元：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.
-
Backbone-based Dynamic Spatio-Temporal Graph Neural Network for epidemic forecasting 査読あり
Junkai Mao, Yuexing Han, Gouhei Tanaka, Bing Wang
Knowledge-Based Systems 296 111952 - 111952 2024年07月
掲載種別：研究論文（学術雑誌）
Accurate epidemic forecasting is a critical task in controlling epidemic spread. Many deep learning-based models focus only on static or dynamic graphs when dealing with spatial information, ignoring their relationship. Additionally, these models often rely on recurrent structures, which can lead to error accumulation and computational time consumption. To address the aforementioned problems, we propose a novel model called Backbone-based Dynamic Spatio-Temporal Graph Neural Network (BDSTGNN). Intuitively, the continuous and smooth changes in graph structure make adjacent graph structures share a basic pattern. To capture this property, we use adaptive methods to generate static backbone graphs containing the primary information, and use temporal models to generate dynamic temporal graphs, and then fuse them to generate a backbone-based dynamic graph. To overcome potential limitations associated with recurrent structures, we introduce a linear model DLinear to handle temporal dependencies, and combine it with dynamic graph convolution for epidemic forecasting. Extensive experiments on two datasets demonstrate that BDSTGNN outperforms baseline models, and ablation comparison further verifies the effectiveness of model components. Furthermore, we analyze and measure the significance of backbone and temporal graphs by using information metrics from different aspects. Finally, we verify the superior efficiency of the BDSTGNN.
-
Designing Network Topologies of Multiple Reservoir Echo State Networks: A Genetic Algorithm Based Approach 査読あり
Ziqiang Li, Kantaro Fujiwara, Gouhei Tanaka
2024 International Joint Conference on Neural Networks (IJCNN) 148 1 - 9 2024年06月
担当区分：最終著者, 責任著者 掲載種別：研究論文（国際会議プロシーディングス） 出版者・発行元：IEEE
-
Reconstructive reservoir computing for anomaly detection in time-series signals 査読あり
Junya Kato, Gouhei Tanaka, Ryosho Nakane, Akira Hirose
Nonlinear Theory and Its Applications, IEICE 15 ( 1 ) 183 - 204 2024年01月
記述言語：英語 掲載種別：研究論文（学術雑誌） 出版者・発行元：Institute of Electronics, Information and Communications Engineers (IEICE)
DOI： 10.1587/nolta.15.183
-
Predicting unobserved climate time series data at distant areas via spatial correlation using reservoir computing.
Shihori Koyama, Daisuke Inoue, Hiroaki Yoshida, Kazuyuki Aihara, Gouhei Tanaka
CoRR abs/2406.03061 2024年
-
Diffusion model for relational inference.
Shuhan Zheng, Ziqiang Li, Kantaro Fujiwara, Gouhei Tanaka
CoRR abs/2401.16755 2024年
-
Dynamical Graph Echo State Networks with Snapshot Merging for Dissemination Process Classification. 査読あり
Ziqiang Li, Kantaro Fujiwara, Gouhei Tanaka
Lecture Notes in Computer Science (LNCS) 2023年11月
-
Dynamical Graph Echo State Networks with Snapshot Merging for Spreading Process Classification 査読あり
Ziqiang Li, Kantaro Fujiwara, Gouhei Tanaka
Communications in Computer and Information Science 1964 CCIS 523 - 534 2023年11月
担当区分：最終著者 掲載種別：研究論文（国際会議プロシーディングス）
The Spreading Process Classification (SPC) is a popular application of temporal graph classification. The aim of SPC is to classify different spreading patterns of information or pestilence within a community represented by discrete-time temporal graphs. Recently, a reservoir computing-based model named Dynamical Graph Echo State Network (DynGESN) has been proposed for processing temporal graphs with relatively high effectiveness and low computational costs. Inspired by DynGESN, we propose a novel reservoir computing-based model called the Grouped Dynamical Graph Echo State Network (GDGESN) for dealing with SPC tasks. In this model, a novel augmentation strategy named the snapshot merging strategy is designed for forming new snapshots by merging neighboring snapshots over time, and then multiple reservoir encoders are set for capturing spatiotemporal features from merged snapshots. After those, the logistic regression is adopted for decoding the sum-pooled embeddings into the classification results. Experimental results on six benchmark SPC datasets show that our proposed model has better classification performances than the DynGESN and several kernel-based models.
DOI： 10.1007/978-981-99-8141-0_39
その他リンク： https://dblp.uni-trier.de/db/conf/iconip/iconip2023-10.html#LiFT23
-
An Echo State Network-Based Method for Identity Recognition with Continuous Blood Pressure Data 査読あり
Ziqiang Li, Kantaro Fujiwara, Gouhei Tanaka
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 14257 LNCS 13 - 25 2023年09月
担当区分：最終著者, 責任著者 記述言語：英語 掲載種別：研究論文（国際会議プロシーディングス）
With the development of Continuous Blood Pressure (CBP) monitoring devices, we can collect real-time blood pressure non-invasively and accurately. Since CBP data can reflect the unique dynamical characteristics of the cardiovascular system for each person, it is reasonable to develop an identity recognition method based on these data. In this study, we propose an Echo State Network-based identity recognition method with CBP data. In the proposed method, we divide each CBP series data into several CBP segments. Then we use a Bi-directional Echo State Network to transform the input segments into high-dimensional reservoir states. Finally, we compute the identity recognition results in an aggregation mode. To evaluate the proposed method, we performed person identification tasks using ten sub-datasets sampled from a large-scale CBP dataset. Our proposed method achieved higher recognition accuracy than other relevant methods in spite of its relatively low computational cost on segment-by-segment and aggregated recognition tasks, respectively.
-
Time-domain Fading Channel Prediction Based on Spin-wave Reservoir Computing. 査読あり
Jiaxuan Chen, Haotian Chen, Ryosho Nakane, Gouhei Tanaka, Akira Hirose
2023 International Joint Conference on Neural Networks 1 - 8 2023年08月
記述言語：英語 掲載種別：研究論文（国際会議プロシーディングス）
-
Film-penetrating transducers applicable to on-chip reservoir computing with spin waves
Jiaxuan Chen, Ryosho Nakane, Gouhei Tanaka, Akira Hirose
Journal of Applied Physics 132 ( 12 ) 2022年09月
掲載種別：研究論文（学術雑誌）
We have proposed a spin-wave transducer structure named film-penetrating transducers (FPTs). FPTs penetrate an on-chip magnetic film for a spin-wave transmission medium and allow flexible spatial arrangements of many exciters/detectors due to their zero-dimensional feature. We constructed four device models with different spatial arrangements of FPT/conventional exciters using a 10-nm-thick ferrimagnetic garnet film with a central FPT detector. We performed numerical experiments that combine electromagnetics with micromagnetics including thermal noise at 300 K. We evaluated important device features of FPTs, such as the signal-to-noise ratios (SNRs), input/output signal transmission efficiencies, and nonlinear phenomena of spin waves. We applied in-phase sinusoidal input currents with various amplitudes and frequencies and altered the damping strengths near the film boundaries. We obtained sufficient SNRs for the practical use of FPTs and revealed that FPTs have both higher transmission efficiencies and nonlinear strengths than conventional antennas, as the input frequency approaches the ferromagnetic resonance frequency of the film. Moreover, we observed and analyzed various nonlinear phenomena of spin waves, including beats in the time-domain waveform, components of integer harmonic frequencies, wide-range scatterings of inter-harmonic frequencies, and frequency doubling in spin precession. These characteristics probably originate from various device effects: FPTs effectively excite dipolar spin waves with large-angle precession, propagating spin waves reflect from the film boundaries, and spin waves dynamically and nonlinearly interfere with each other. This study demonstrated that FPTs have promising features for both their applications to reservoir computing and the studies on the physics of nonlinear and space-varying spin waves.
DOI： 10.1063/5.0102974
前のページ - 