Hideaki Hayashi

Hideaki Hayashi

Auto-generated profile, based on publicly available information

Hideaki Hayashi

D3 Center, The University of Osaka, Suita, Japan

Research Focus & Metrics

Academic Career Length

14Years (2013-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
15 October 2019 in IEEE Access
Verified
Journal Article
Open access
Changhee HanChanghee Han
Changhee Han
Leonardo RundoLeonardo Rundo
Leonardo Rundo
RARyosuke Araki
Ryosuke Araki
YNYudai Nagano
Yudai Nagano
YFYujiro Furukawa
Yujiro Furukawa
HHHideaki Hayashi
Hideaki Hayashi

12 August 2019 in IEEE Access
Verified
Journal Article
Open access
SHShota Harada
Shota Harada
HHHideaki Hayashi
Hideaki Hayashi
SUSeiichi Uchida
Seiichi Uchida
The effectiveness of biosignal generation and data augmentation with biosignal generative models based on generative adversarial networks (GANs), which are a type of deep learning technique, was demonstrated in our previous paper. GAN-based generative models only learn the projection between a random distribution as input data and the distribution of training data. Therefore, the relationship between input and generated data is unclear, and the characteristics of the data generated from this model cannot be controlled. This study proposes a method for generating time-series data based on GANs and explores their ability to generate biosignals with certain classes and characteristics. Moreover, in the proposed method, latent variables are analyzed using canonical correlation analysis (CCA) to represent the relationship between input and generated data as canonical loadings. Using these loadings, we can control the characteristics of the data generated by the proposed method. The influence of class labels on generated data is analyzed by feeding the data interpolated between two class labels into the generator of the proposed GANs. The CCA of the latent variables is shown to be an effective method of controlling the generated data characteristics. We are able to model the distribution of the time-series data without requiring domain-dependent knowledge using the proposed method. Furthermore, it is possible to control the characteristics of these data by analyzing the model trained using the proposed method. To the best of our knowledge, this work is the first to generate biosignals using GANs while controlling the characteristics of the generated data.

25 January 2017 in IEEE Transactions on Biomedical Engineering
Verified
Journal Article
Closed access
HHHideaki Hayashi
Hideaki Hayashi
AFAkira Furui
Akira Furui
Yuichi KuritaYuichi Kurita
Yuichi Kurita
TTToshio Tsuji
Toshio Tsuji
Objective: This paper describes the formulation of a surface electromyogram (EMG) model capable of representing the variance distribution of EMG signals. Methods: In the model, EMG signals are handled based on a Gaussian white noise process with a mean of zero for each variance value. EMG signal variance is taken as a random variable that follows inverse gamma distribution, allowing the representation of noise superimposed onto this variance. Variance distribution estimation based on marginal likelihood maximization is also outlined in this paper. The procedure can be approximated using rectified and smoothed EMG signals, thereby allowing the determination of distribution parameters in real time at low computational cost. Results: A simulation experiment was performed to evaluate the accuracy of distribution estimation using artificially generated EMG signals, with results demonstrating that the proposed model's accuracy is higher than that of maximum-likelihood-based estimation. Analysis of variance distribution using real EMG data also suggested a relationship between variance distribution and signal-dependent noise. Conclusion: The study reported here was conducted to examine the performance of a proposed surface EMG model capable of representing variance distribution and a related distribution parameter estimation method. Experiments using artificial and real EMG data demonstrated the validity of the model. Significance: Variance distribution estimated using the proposed model exhibits potential in the estimation of muscle force.

19 February 2015 in IEEE Transactions on Neural Networks and Learning Systems
Verified
Journal Article
Closed access
HHHideaki Hayashi
Hideaki Hayashi
TSTaro Shibanoki
Taro Shibanoki
KSKeisuke Shima
Keisuke Shima
YKYuichi Kurita
Yuichi Kurita
TTToshio Tsuji
Toshio Tsuji
This paper proposes a probabilistic neural network (NN) developed on the basis of time-series discriminant component analysis (TSDCA) that can be used to classify high-dimensional time-series patterns. TSDCA involves the compression of high-dimensional time series into a lower dimensional space using a set of orthogonal transformations and the calculation of posterior probabilities based on a continuous-density hidden Markov model with a Gaussian mixture model expressed in the reduced-dimensional space. The analysis can be incorporated into an NN, which is named a time-series discriminant component network (TSDCN), so that parameters of dimensionality reduction and classification can be obtained simultaneously as network coefficients according to a backpropagation through time-based learning algorithm with the Lagrange multiplier method. The TSDCN is considered to enable high-accuracy classification of high-dimensional time-series patterns and to reduce the computation time taken for network training. The validity of the TSDCN is demonstrated for high-dimensional artificial data and electroencephalogram signals in the experiments conducted during the study.

28 January 2019 in IEEE Transactions on Biomedical Engineering
Verified
Journal Article
Closed access
AFAkira Furui
Akira Furui
HHHideaki Hayashi
Hideaki Hayashi
TTToshio Tsuji
Toshio Tsuji
Objective: Surface electromyogram (EMG) signals have typically been assumed to follow a Gaussian distribution. However, the presence of non-Gaussian signals associated with muscle activity has been reported in recent studies, and there is no general model of the distribution of EMG signals that can explain both non-Gaussian and Gaussian distributions within a unified scheme. Methods: In this paper, we describe the formulation of a non-Gaussian EMG model based on a scale mixture distribution. In the model, an EMG signal at a certain time follows a Gaussian distribution, and its variance is handled as a random variable that follows an inverse gamma distribution. Accordingly, the probability distribution of EMG signals is assumed to be a mixture of Gaussians with the same mean but different variances. The EMG variance distribution is estimated via marginal likelihood maximization. Results: Experiments involving nine participants revealed that the proposed model provides a better fit to recorded EMG signals than conventional EMG models. It was also shown that variance distribution parameters may reflect underlying motor unit activity. Conclusion: This study proposed a scale mixture distribution-based stochastic EMG model capable of representing changes in non-Gaussianity associated with muscle activity. A series of experiments demonstrated the validity of the model and highlighted the relationship between the variance distribution and muscle force. Significance: The proposed model helps to clarify conventional wisdom regarding the probability distribution of surface EMG signals within a unified scheme.

6 February 2024 in IEEE Transactions on Neural Networks and Learning Systems
Verified
Journal Article
Closed access
HHHideaki Hayashi
Hideaki Hayashi
Generative models offer advantageous characteristics for classification tasks, such as the availability of unsupervised data and calibrated confidence. In contrast, discriminative models have advantages in terms of their potential to outperform their generative counterparts and the simplicity of their model structures and learning algorithms. In this article, we propose a method to train a hybrid of discriminative and generative models in a single neural network (NN), which exhibits the characteristics of both models. The key idea is the Gaussian-coupled softmax layer, which is a fully connected layer with a softmax activation function coupled with Gaussian distributions. This layer can be embedded into an NN-based classifier and allows the classifier to estimate both the class posterior distribution and the input data distribution. We demonstrate that the proposed hybrid model can be applied to semi-supervised learning and confidence calibration.