Yuichiro Tanaka
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Yuichiro Tanaka has a total of 62 co-authors. Here are the top co-authors by number of shared publications:
- Hakaru Tamukoh
Grad. Sch. of Life Sci. and Syst. Eng., Kyushu Inst. of Tech
13 shared publications - Hirofumi Tanaka
Department of Human Intelligence Systems Graduate School of Life Science and Systems Engineering Kyushu Institute of Technology Kitakyushu Japan
5 shared publications - Yuki Usami
Department of Human Intelligence Systems Graduate School of Life Science and Systems Engineering Kyushu Institute of Technology Kitakyushu Japan
5 shared publications - Arie Rachmad Syulistyo
Graduate School of Life Science and Systems Engineering Kyushu Institute of Technology 2-4 Hibikino Wakamatsu 808-0196 Kitakyushu Japan
3 shared publications Chieko Shimada
Department of Geology and Paleontology, National Museum of Nature and Science, 1-4 Amakubo, Tsukuba-shi, Ibaraki 3050005, Japan ( )
3 shared publications- Nitin Kumar Singh
Graduate School of Life Science and Systems Engineering Kyushu Institute of Technology 2-4 Hibikino Wakamatsu 808-0196 Kitakyushu Japan
3 shared publications
Latest Publications
- Open Access
- 21 September 2026
- Published by IOS Press
- p. 949-961
Blink Detection with an Event-Based Vision Sensor and an Echo State Network Under In-Vehicle Background Noise
In this study, we propose a blink detection method for in-vehicle background-noise conditions using an event-based vision sensor (EVS) and evaluate its effectiveness. Existing EVS-based and template-matching methods are vulnerable to background noise. In this study, we construct a method that feeds activity waveforms computed from event data into an echo state network (ESN) to detect blink segments in time series under in-vehicle background-noise conditions. Using a simple EVS dataset that combines indoor blink sequences and in-vehicle background-noise sequences, we compare the proposed ESN-based detector with a conventional template-matching baseline under the constraint that all true blinks are detected. We also analyze how the number of training blinks affects the overall detection performance. Experiments show that, in this simplified setting, our method maintains full blink-detection recall while substantially reducing false positives compared with template matching. Our experiments also provide initial insight into the minimum number of training blinks required for reliable blink detection. Extending the evaluation to continuous in-vehicle driving data with more diverse blink patterns, drivers, and vehicles is left as future work.
- 21 September 2026
- Published by IOS Press
- p. 949-961
Open AccessBlink Detection with an Event-Based Vision Sensor and an Echo State Network Under In-Vehicle Background Noise
In this study, we propose a blink detection method for in-vehicle background-noise conditions using an event-based vision sensor (EVS) and evaluate its effectiveness. Existing EVS-based and template-matching methods are vulnerable to background noise. In this study, we construct a method that feeds activity waveforms computed from event data into an echo state network (ESN) to detect blink segments in time series under in-vehicle background-noise conditions. Using a simple EVS dataset that combines indoor blink sequences and in-vehicle background-noise sequences, we compare the proposed ESN-based detector with a conventional template-matching baseline under the constraint that all true blinks are detected. We also analyze how the number of training blinks affects the overall detection performance. Experiments show that, in this simplified setting, our method maintains full blink-detection recall while substantially reducing false positives compared with template matching. Our experiments also provide initial insight into the minimum number of training blinks required for reliable blink detection. Extending the evaluation to continuous in-vehicle driving data with more diverse blink patterns, drivers, and vehicles is left as future work.
- Open Access
- 1 January 2026
- Published by Institute of Electronics, Information and Communications Engineers (IEICE) in Nonlinear Theory and Its Applications, IEICE
- Vol. 17 (1) , 108-123
Environment-specific knowledge acquisition system for home service robots via brain-inspired memory and commonsense knowledge
Home-service robots are expected to enhance the quality of daily life. Two types of knowledge, commonsense and environment-specific knowledge, are required for home-service robots. Current robots can handle audio-visual-based common knowledge, task planning, and action generation. Additionally, several systems that handle environment-specific knowledge have been developed. Previously proposed brain-inspired models can integrate visual and location information to represent episodes; however, it is difficult to obtain novel environment-specific knowledge. In this study, we propose a system that efficiently acquires novel environment-specific knowledge by combining environment-specific knowledge stored in a brain-inspired memory system with commonsense knowledge inferred by large language models (LLMs). We verified the performance of common-sense retrieval from LLMs and evaluated the effectiveness of combining environment-specific knowledge and commonsense knowledge in the home environment.
- 1 January 2026
- Published by Institute of Electronics, Information and Communications Engineers (IEICE) in Nonlinear Theory and Its Applications, IEICE
- Vol. 17 (1) , 108-123
Open AccessEnvironment-specific knowledge acquisition system for home service robots via brain-inspired memory and commonsense knowledge
Home-service robots are expected to enhance the quality of daily life. Two types of knowledge, commonsense and environment-specific knowledge, are required for home-service robots. Current robots can handle audio-visual-based common knowledge, task planning, and action generation. Additionally, several systems that handle environment-specific knowledge have been developed. Previously proposed brain-inspired models can integrate visual and location information to represent episodes; however, it is difficult to obtain novel environment-specific knowledge. In this study, we propose a system that efficiently acquires novel environment-specific knowledge by combining environment-specific knowledge stored in a brain-inspired memory system with commonsense knowledge inferred by large language models (LLMs). We verified the performance of common-sense retrieval from LLMs and evaluated the effectiveness of combining environment-specific knowledge and commonsense knowledge in the home environment.
- Open Access
- 1 January 2026
- Published by Institute of Electronics, Information and Communications Engineers (IEICE) in Nonlinear Theory and Its Applications, IEICE
- Vol. 17 (1) , 79-92
Sign language recognition using parallel bidirectional reservoir computing
Sign language recognition (SLR) facilitates communication between deaf and hearing communities. Deep learning based SLR models are commonly used but require extensive computational resources, making them unsuitable for deployment on edge devices. To address these limitations, we propose a lightweight SLR system that combines parallel bidirectional reservoir computing (PBRC) with MediaPipe. MediaPipe enables real-time hand tracking and precise extraction of hand joint coordinates, which serve as input features for the PBRC architecture. The proposed PBRC architecture consists of two echo state network (ESN) based bidirectional reservoir computing (BRC) modules arranged in parallel to capture temporal dependencies, thereby creating a rich feature representation for classification. We trained our PBRC-based SLR system on the Word-Level American Sign Language (WLASL) video dataset, achieving top-1, top-5, and top-10 accuracies of 60.85%, 85.86%, and 91.74%, respectively. Training time was significantly reduced to 18.67 seconds due to the intrinsic properties of reservoir computing, compared to over 55 minutes for deep learning based methods such as Bi-GRU. This approach offers a lightweight, cost-effective solution for real-time SLR on edge devices.
- 1 January 2026
- Published by Institute of Electronics, Information and Communications Engineers (IEICE) in Nonlinear Theory and Its Applications, IEICE
- Vol. 17 (1) , 79-92
Open AccessSign language recognition using parallel bidirectional reservoir computing
Sign language recognition (SLR) facilitates communication between deaf and hearing communities. Deep learning based SLR models are commonly used but require extensive computational resources, making them unsuitable for deployment on edge devices. To address these limitations, we propose a lightweight SLR system that combines parallel bidirectional reservoir computing (PBRC) with MediaPipe. MediaPipe enables real-time hand tracking and precise extraction of hand joint coordinates, which serve as input features for the PBRC architecture. The proposed PBRC architecture consists of two echo state network (ESN) based bidirectional reservoir computing (BRC) modules arranged in parallel to capture temporal dependencies, thereby creating a rich feature representation for classification. We trained our PBRC-based SLR system on the Word-Level American Sign Language (WLASL) video dataset, achieving top-1, top-5, and top-10 accuracies of 60.85%, 85.86%, and 91.74%, respectively. Training time was significantly reduced to 18.67 seconds due to the intrinsic properties of reservoir computing, compared to over 55 minutes for deep learning based methods such as Bi-GRU. This approach offers a lightweight, cost-effective solution for real-time SLR on edge devices.
- Open Access
- 1 January 2026
- Published by Institute of Electronics, Information and Communications Engineers (IEICE) in Nonlinear Theory and Its Applications, IEICE
- Vol. 17 (1) , 39-65
Effective simulation of in-materio physical reservoirs: A unidirectional cluster-based echo state network approach
Fabricated in-materio reservoir computing devices have demonstrated ultra-low power consumption for unconventional computing tasks across varieties of material systems. Simulating their internal networks is desired to analyze them, but challenging due to physical variability. While echo state networks (ESNs) provide a relevant framework, conventional ESN simulations require a long computational time at large scales and fail to replicate physical device outputs. We propose a novel ESN framework incorporating structural constraints inspired by physical in-materio devices, featuring a unidirectional cluster-based topology. Our approach demonstrates high consistency, faster computation, and promising benchmark performance compared to conventional ESNs.
- 1 January 2026
- Published by Institute of Electronics, Information and Communications Engineers (IEICE) in Nonlinear Theory and Its Applications, IEICE
- Vol. 17 (1) , 39-65
Open AccessEffective simulation of in-materio physical reservoirs: A unidirectional cluster-based echo state network approach
Fabricated in-materio reservoir computing devices have demonstrated ultra-low power consumption for unconventional computing tasks across varieties of material systems. Simulating their internal networks is desired to analyze them, but challenging due to physical variability. While echo state networks (ESNs) provide a relevant framework, conventional ESN simulations require a long computational time at large scales and fail to replicate physical device outputs. We propose a novel ESN framework incorporating structural constraints inspired by physical in-materio devices, featuring a unidirectional cluster-based topology. Our approach demonstrates high consistency, faster computation, and promising benchmark performance compared to conventional ESNs.
- Open Access
Sign Language Recognition using Parallel Bidirectional Reservoir Computing
Sign language recognition (SLR) facilitates communication between deaf and hearing communities. Deep learning based SLR models are commonly used but require extensive computational resources, making them unsuitable for deployment on edge devices. To address these limitations, we propose a lightweight SLR system that combines parallel bidirectional reservoir computing (PBRC) with MediaPipe. MediaPipe enables real-time hand tracking and precise extraction of hand joint coordinates, which serve as input features for the PBRC architecture. The proposed PBRC architecture consists of two echo state network (ESN) based bidirectional reservoir computing (BRC) modules arranged in parallel to capture temporal dependencies, thereby creating a rich feature representation for classification. We trained our PBRC-based SLR system on the Word-Level American Sign Language (WLASL) video dataset, achieving top-1, top-5, and top-10 accuracies of 60.85%, 85.86%, and 91.74%, respectively. Training time was significantly reduced to 18.67 seconds due to the intrinsic properties of reservoir computing, compared to over 55 minutes for deep learning based methods such as Bi-GRU. This approach offers a lightweight, cost-effective solution for real-time SLR on edge devices.
Open AccessSign Language Recognition using Parallel Bidirectional Reservoir Computing
Sign language recognition (SLR) facilitates communication between deaf and hearing communities. Deep learning based SLR models are commonly used but require extensive computational resources, making them unsuitable for deployment on edge devices. To address these limitations, we propose a lightweight SLR system that combines parallel bidirectional reservoir computing (PBRC) with MediaPipe. MediaPipe enables real-time hand tracking and precise extraction of hand joint coordinates, which serve as input features for the PBRC architecture. The proposed PBRC architecture consists of two echo state network (ESN) based bidirectional reservoir computing (BRC) modules arranged in parallel to capture temporal dependencies, thereby creating a rich feature representation for classification. We trained our PBRC-based SLR system on the Word-Level American Sign Language (WLASL) video dataset, achieving top-1, top-5, and top-10 accuracies of 60.85%, 85.86%, and 91.74%, respectively. Training time was significantly reduced to 18.67 seconds due to the intrinsic properties of reservoir computing, compared to over 55 minutes for deep learning based methods such as Bi-GRU. This approach offers a lightweight, cost-effective solution for real-time SLR on edge devices.
- Open Access
Sign Language Recognition using Bidirectional Reservoir Computing
Sign language recognition (SLR) facilitates communication between deaf and hearing individuals. Deep learning is widely used to develop SLR-based systems; however, it is computationally intensive and requires substantial computational resources, making it unsuitable for resource-constrained devices. To address this, we propose an efficient sign language recognition system using MediaPipe and an echo state network (ESN)-based bidirectional reservoir computing (BRC) architecture. MediaPipe extracts hand joint coordinates, which serve as inputs to the ESN-based BRC architecture. The BRC processes these features in both forward and backward directions, efficiently capturing temporal dependencies. The resulting states of BRC are concatenated to form a robust representation for classification. We evaluated our method on the Word-Level American Sign Language (WLASL) video dataset, achieving a competitive accuracy of 57.71% and a significantly lower training time of only 9 seconds, in contrast to the 55 minutes and $38$ seconds required by the deep learning-based Bi-GRU approach. Consequently, the BRC-based SLR system is well-suited for edge devices.Open AccessSign Language Recognition using Bidirectional Reservoir Computing
Sign language recognition (SLR) facilitates communication between deaf and hearing individuals. Deep learning is widely used to develop SLR-based systems; however, it is computationally intensive and requires substantial computational resources, making it unsuitable for resource-constrained devices. To address this, we propose an efficient sign language recognition system using MediaPipe and an echo state network (ESN)-based bidirectional reservoir computing (BRC) architecture. MediaPipe extracts hand joint coordinates, which serve as inputs to the ESN-based BRC architecture. The BRC processes these features in both forward and backward directions, efficiently capturing temporal dependencies. The resulting states of BRC are concatenated to form a robust representation for classification. We evaluated our method on the Word-Level American Sign Language (WLASL) video dataset, achieving a competitive accuracy of 57.71% and a significantly lower training time of only 9 seconds, in contrast to the 55 minutes and $38$ seconds required by the deep learning-based Bi-GRU approach. Consequently, the BRC-based SLR system is well-suited for edge devices. - Open Access
- 27 November 2024
- Published by Wiley in Advanced Electronic Materials
- Vol. 10 (12)
Thermally Stable Ag2Se Nanowire Network as an Effective In‐Materio Physical Reservoir Computing Device
The artificial intelligence (AI) paradigm shifts from software to implementing general‐purpose or application‐specific hardware systems with lower power requirements. This study explored a material physical reservoir consisting of a material random network, called in‐materio physical reservoir computing (RC) to achieve efficient hardware systems. The device, made up of a random, highly interconnected network of nonlinear Ag2Se nanojunctions as reservoir nodes, demonstrated the requisite characteristics of an in‐materio physical reservoir, including but not limited to nonlinear switching, memory, and higher harmonic generation. The power consumption of the in‐materio physical reservoir is 0.07 nW per nanojunctions, confirming its highly efficient information processing system. As a hardware reservoir, the devices successfully performed waveform generation tasks. Finally, a voice classification by an in‐materio physical reservoir is achieved over 80%, comparable to an RC software simulation. In‐materio physical RC with rich nonlinear dynamics has huge potential for next‐generation hardware‐based AI.
- 27 November 2024
- Published by Wiley in Advanced Electronic Materials
- Vol. 10 (12)
Open AccessThermally Stable Ag2Se Nanowire Network as an Effective In‐Materio Physical Reservoir Computing Device
The artificial intelligence (AI) paradigm shifts from software to implementing general‐purpose or application‐specific hardware systems with lower power requirements. This study explored a material physical reservoir consisting of a material random network, called in‐materio physical reservoir computing (RC) to achieve efficient hardware systems. The device, made up of a random, highly interconnected network of nonlinear Ag2Se nanojunctions as reservoir nodes, demonstrated the requisite characteristics of an in‐materio physical reservoir, including but not limited to nonlinear switching, memory, and higher harmonic generation. The power consumption of the in‐materio physical reservoir is 0.07 nW per nanojunctions, confirming its highly efficient information processing system. As a hardware reservoir, the devices successfully performed waveform generation tasks. Finally, a voice classification by an in‐materio physical reservoir is achieved over 80%, comparable to an RC software simulation. In‐materio physical RC with rich nonlinear dynamics has huge potential for next‐generation hardware‐based AI.
- Open Access
- 1 January 2024
- Published by Institute of Electronics, Information and Communications Engineers (IEICE) in Nonlinear Theory and Its Applications, IEICE
- Vol. 15 (4) , 883-898
Object recognition and grasping point detection using carbon nanotube - polydimethylsiloxane nanocomposite sensor
This paper presents a system for processing tactile information using a carbon nanotube (CNT)-polydimethylsiloxane (PDMS) nanocomposite sensor designed for robotic applications. This study introduces an approach for recognizing objects and detecting optimal grasping points using tactile data from a sensor-equipped robotic hand. The sensor is expected to be more efficient than a computerized implementation using sensor dynamics directly in the computation. The experiments demonstrated the system's ability to classify nine types of objects with an accuracy of 83.56% and to discern two grasping points on four object types, achieving a 71.7% success rate.
- 1 January 2024
- Published by Institute of Electronics, Information and Communications Engineers (IEICE) in Nonlinear Theory and Its Applications, IEICE
- Vol. 15 (4) , 883-898
Open AccessObject recognition and grasping point detection using carbon nanotube - polydimethylsiloxane nanocomposite sensor
This paper presents a system for processing tactile information using a carbon nanotube (CNT)-polydimethylsiloxane (PDMS) nanocomposite sensor designed for robotic applications. This study introduces an approach for recognizing objects and detecting optimal grasping points using tactile data from a sensor-equipped robotic hand. The sensor is expected to be more efficient than a computerized implementation using sensor dynamics directly in the computation. The experiments demonstrated the system's ability to classify nine types of objects with an accuracy of 83.56% and to discern two grasping points on four object types, achieving a 71.7% success rate.
- Open Access
- 7 December 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 11, 138839-138849
Self-Organizing Multiple Readouts for Reservoir Computing
With advancements in deep learning (DL), artificial intelligence (AI) technology has become an indispensable tool. However, the application of DL incurs significant computational costs, making it less viable for edge AI scenarios. Consequently, the demand for cost-effective AI solutions, other than DL-based approaches, is increasing. Reservoir computing (RC) has attracted interest owing to its ability to provide low-cost training alternatives, holding great promise for edge AI applications. However, the training capability of RC is constrained by its reliance on a single linear layer, while weight connections in the remaining layers remain static during training. Moreover, accomplishing continuous learning tasks is difficult owing to the catastrophic forgetting in the linear layer. Therefore, we propose the integration of self-organizing multiple readouts to enhance RC’s training capability. Our method distributes training data across multiple readouts, which prevents catastrophic forgetting of readouts and empowers each readout to adeptly assimilate new data, thereby elevating the overall training performance. The self-organizing function, which assigns similar data to the same readout, optimizes the memory utilization of these multiple readouts. Experimental results show that an RC equipped with the proposed multiple readouts successfully solved a continuous learning task by mitigating catastrophic forgetting because of the data distribution to the multiple readouts. Additionally, the RC achieved higher accuracy in a sound recognition task compared with the existing RC paradigm because of ensemble learning in the multiple readouts. Multiple readouts are effective in enhancing the training capability of RC and can contribute to the realization of RC applications.
- 7 December 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 11, 138839-138849
Open AccessSelf-Organizing Multiple Readouts for Reservoir Computing
With advancements in deep learning (DL), artificial intelligence (AI) technology has become an indispensable tool. However, the application of DL incurs significant computational costs, making it less viable for edge AI scenarios. Consequently, the demand for cost-effective AI solutions, other than DL-based approaches, is increasing. Reservoir computing (RC) has attracted interest owing to its ability to provide low-cost training alternatives, holding great promise for edge AI applications. However, the training capability of RC is constrained by its reliance on a single linear layer, while weight connections in the remaining layers remain static during training. Moreover, accomplishing continuous learning tasks is difficult owing to the catastrophic forgetting in the linear layer. Therefore, we propose the integration of self-organizing multiple readouts to enhance RC’s training capability. Our method distributes training data across multiple readouts, which prevents catastrophic forgetting of readouts and empowers each readout to adeptly assimilate new data, thereby elevating the overall training performance. The self-organizing function, which assigns similar data to the same readout, optimizes the memory utilization of these multiple readouts. Experimental results show that an RC equipped with the proposed multiple readouts successfully solved a continuous learning task by mitigating catastrophic forgetting because of the data distribution to the multiple readouts. Additionally, the RC achieved higher accuracy in a sound recognition task compared with the existing RC paradigm because of ensemble learning in the multiple readouts. Multiple readouts are effective in enhancing the training capability of RC and can contribute to the realization of RC applications.
- Open Access
- 1 January 2023
- Published by IOP Publishing in Applied Physics Express
- Vol. 16 (1) , 014002
Random network device fabricated using Ag2Se nanowires for data augmentation with binarized convolutional neural network
An Ag2Se nanowire random network was fabricated for application as a data augmentation device and combined with a binary convolutional neural network (BCNN) to achieve high accuracy in voice classification tasks. Due to the nonlinear high-dimensional characteristics resulting from the formation of the conductive filament at the cross junction, the Ag2Se device could transform input data into higher-order multiple signals, thereby enhancing the accuracy of the classification task by augmenting input signals. The results indicate that materials can realize data augmentation with the same performance as software, suggesting that material-based hardware can be used as an elemental technology for information processing.
- 1 January 2023
- Published by IOP Publishing in Applied Physics Express
- Vol. 16 (1) , 014002
Open AccessRandom network device fabricated using Ag2Se nanowires for data augmentation with binarized convolutional neural network
An Ag2Se nanowire random network was fabricated for application as a data augmentation device and combined with a binary convolutional neural network (BCNN) to achieve high accuracy in voice classification tasks. Due to the nonlinear high-dimensional characteristics resulting from the formation of the conductive filament at the cross junction, the Ag2Se device could transform input data into higher-order multiple signals, thereby enhancing the accuracy of the classification task by augmenting input signals. The results indicate that materials can realize data augmentation with the same performance as software, suggesting that material-based hardware can be used as an elemental technology for information processing.