We had proposed a stochastic (or statistical) Equivalent linearization - Gaussian Sum Filter (abbreviated as EqGS Filter) for discrete time nonlinear Systems. Subsequently, in this paper, we investigate and show the further results related to the EqGS Filter. Especially we discuss a method to apply Gauss-Hermite quadrature rules for evaluation of the conditional expected values of the quantities required to design the EqGS filter.

View full abstract

This paper treats the filtering and parameter identification for the stochastic diffusion systems with unknown boundary conditions. The physical situation of the unknown boundary conditions can be found in many industrial problems,i.g., the salt concentration model of the river Rhine is a typical example . After formulating the diffusion systems by regarding the noisy observation data near the systems boundary region as the system's boundary inputs, we derive the Kalman filter and the related likelihood function. Some numerical examples are demonstrated.

View full abstract

In this paper, the author investigates the statistical moments such as expectations and variances for a class of fuzzy random sets, where the fuzzy random set is considered as a model of the capricious vague perception of a crisp phenomenon or a crisp random phenomenon. First, the class of fuzzy random sets, which has been proposed by author[1, 2], where the vague perception of a crisp phenomenon fluctuates slightly but randomly by the state of a capricious person's mind, is refined and its expectation and variance are introduced. Secondly, the refined class of fuzzy random sets is extended to the models for the capricious vague perceptions of crisp random phenomena, and their expectations and variances are investigated from the viewpoint of the multivalued logic.

View full abstract

In this paper, we are concerned with a problem of optimization of the linear observations which are used in the stationary Kalman filter. Especially, we consider the optimization of the gain matrix in the observation. In the previous works one of the authors, an information theoretic criterion, based on a generalized Water Filling Theorem, was introduced to obtain a gain matrix which minimizes the stationary error variance. The merit of this approach is that both analytical and numerical solutions are rather easily obtained compared with the case of the performance criterion which is quadratic in the estimation error and the gain matrix. In this solution process, however, the Riccati equation of the error covariance matrix reduces to a quasi linear equation, and the condition for the existence of the solution of this equation is somewhat stronger than that of the usual Riccati equation. This paper is concerned with the case of the quadratic performance criterion. We propose a new method of numerical optimization by introducing the gradient method and a new rule of updating the angular parameters which are brought by the polar-coordinate representation of an orthogonal matrix. The results of numerical experiments show the efficiency of the algorithm.

View full abstract

According the competitive environment of the global economy, many entrepreneurs become to pay more attention to the knowledge. They want to Enhance competitiveness by developing the knowledge level. Considering the knowledge factor, we built the simulation model and do the simulation to find the effect of the knowledge to enterprise.

View full abstract

We can utter sounds by changing the shape of our vocal tract. The vocal tract characteristic has been analyzed by AR model for speech signals and the vocal tract area function can be estimated by PARCOR coefficients computed from the AR model. Phonological and personal information are included in the vocal tract area function. From this fact, the vocal tract area function can be considered to be available to voice synthesis or speaker recognition. In this paper, we will analyze the vocal tract area with the aim of recognition of phoneme of Japanese vowels.

View full abstract

We consider the dynamical traps model of human fuzzy rationality which describes the behavior of human controling a dynamical system near an equilibrium point. The basic dynamical trap model describes the behavior of human operator neglecting small deviations from the equilubrium point. We propose the extended model that takes into account the effect of imperfect implementation of the desired control strategy. The results of numerical simulations confirm that human fuzzy rationality could be responsible for anomalous behavior of human-controlled systems.

View full abstract

The author has been developing a safe driving system of mobile objects such as silver cars, mobility scooters and so on by using depth sensors which can capture range data of VGA size. If the geometrical relation between the sensor and the space is completely known, each point of the captured range data can be classified into three groups: upper than the ground, on the ground, and lower than the ground. However, it is essential to be able to deal with the unpredictable change of the posture of the sensor due to the movement of the mobile vehicle. The author developed an online estimation scheme of the posture including pitch angle, roll angle, and height from the observed data in the framework of optimization. In this paper, the author proposes an estimation scheme based on the state space model and apply Extended Kalman Filter for the same application problem. By comparing them, we will compare the algorithms by experimental results.

View full abstract

In this paper we study hybrid estimation for linear discrete-time systems with noises not to be restricted to be Gaussian. It is assumed that modes of the systems are not directly accessible. We consider optimal estimation problems to find both estimated states of the systems and a candidate of the distributions of the modes over the finite time interval. We adopt most probable trajectory (MPT) approach. Q. Zhang (1999, 2000) has presented hybrid filtering algorithm, i.e., causal estimation, by MPT approach for linear continuous- and discrete-time hybrid systems with non-Gaussian noises. We consider both filtering and smoothing problems for the linear discrete-time hybrid systems in this paper. Based on the principles of hybrid optimality we present filtering and smoothing algorithms, which give the solutions of these estimation problems. In the smoothing case, we can expect better estimation performance by taking into consideration noncausal information of observations. The hybrid smoother is realized by two filters approach ([23]).

View full abstract

This paper describes the state space model for estimation of the mobile wireless client based on sensor data fusion. Data pertaining to the physical positions of personal electronic devices or mobile robots are important for information services and robotic applications. We deal strength of received signals of wireless LAN signals from the wireless LAN access points for localization of mobile clients in multistory buildings. The output of accelerometer has uncertainty; we apply the probabilistic model of the estimation result of location using wireless LAN signals and the acceleration acquired by the sensor as the observation model. We use an acceleration of the mobile client as the sensor data. In addition, a method for simplified measurement for making the fingerprinting-type estimation system based on the pacing motion by the human is discussed. Experimental results show the feasibility of the method.

View full abstract

In this paper, a new direct controller parameter tuning method is proposed by applying particle filtering to a system derived from the performance index of fictitious reference iterative tuning which is a kind of direct controller tuning method. First, a linear plant is dealt with as a basic case, and then, the proposed method is extended to a plant with a dead-zone property. Two numerical examples are provided to illustrate the effectiveness of the proposed method.

View full abstract

In this paper, the investigation of the ionospheric models for GNSS (Global Navigation Satellite System) in local area like Japan is presented. The ionospheric delay is modeled applying SCHA (Spherical Cap Harmonic Analysis). It is well known that the ionosphere varies according to solar activity. Therefore, ionospheric effect on GNSS positioning varies with not only the position but also the local time, and it has a periodicity. The purpose of this research is to investigate the periodic characteristics of the ionosphere are investigated by utilizing the SCHA model. Furthermore, based on the above investigation, the capability for ionospheric model prediction is discussed.

View full abstract

In this paper, the applications of the Gaussian sum filtering method to the carrier phase based relative GNSS positioning are considered. The distribution of the observation noise in the double differenced carrier phases is modeled by the weighted sum of Gaussian distributions, and the Gaussian sum filter derived by extending and modifying the Gaussian sum filter proposed by Alspach and Sorenson [1] is applied to the positioning. The experiments are done by using real receiver data, and the performance of positioning as well as integer ambiguity estimation are investigated.

View full abstract

In car-navigation systems, car position measurement estimates the car's position and heading angle from the outputs of a GPS receiver and sensors, and calculates the position on a link that represents a road and is recorded in a digital road map. We focus on the latter process, which is map matching. If there are large errors in the car's estimated position and heading angle and in link position and direction angle, map matching selects an incorrect link. If map matching uses the data of the past route, it can select a correct link when the car is on a parallel road. This paper proposes a map-matching method using a chi-squared statistic for car-navigation systems. This method calculates a selection criterion that takes into account the errors in the car's estimated position and heading angle and in the link position and direction angle. This criterion is chi-squared distributed and is calculated from data points that were observed at constant distance intervals. The method's performance was experimentally evaluated using observation data collected in a car. It was found to be effective on roads that pose map matching difficulties, i.e., forks in roads, parallel roads, and roundabouts.

View full abstract

In this paper, we present blind image restoration algorithms for motion blur caused by a motion between the imaging equipment and the object during the exposure time. We explain the motion degradation analysis and how to estimate parameters using Cepstral analysis, Radon transform and Hough transform.

View full abstract

In this paper, preamble design for estimation of frequency selective channels, carrier frequency offset (CFO) and in-phase/quadrature-phase (I/Q) imbalance in orthogonal frequency division multiplexing (OFDM) systems is proposed. First we utilize convex optimization to optimize power of all active subcarriers, then we employ adaptive Markov chain Monte Carlo (AMCMC) techniques to select preamble sequence that minimizes the channel estimate mean squared error (MSE) while suppressing the effect of the I/Q mismatch. To estimate CFO, maximum likelihood (ML) based scheme that utilizes two successive OFDM preambles is employed. The CFO is estimated by considering the phase rotation between two consecutive received OFDM preambles. Numerical simulations are provided to verify the efficacy of the proposed design.

View full abstract

In this paper, a new identification method of discrete-time linear stochastic systems is proposed. We assume that some entries of the system matrix are unknown and propose a new method which identifies those unknown entries and the state vector of the system simultaneously. The key idea of the proposed method is the use of pseudomeasurement which is a fictitious observation process on the unknown entries and has been introduced by Kameyama and Ohsumi for continuous-time linear stochastic systems. Augmenting the pseudomeasurement with the original observation process, we derive the new identification method by applying the extended Kalman filter.

View full abstract

An approach is proposed to identify unknown system parameters which are known a priori to be constrained by their own upper and lower bounds. The inequality constraints are incorporated in the original observation data as pseudomeasurement. The identification is performed within the Kalman filtering framework. Several numerical experiments are provided.

View full abstract

In this paper, for the purpose of evaluating the degradation level of the performance in application with the parametric model in which the estimated parameters have the disturbance, we consider the relation between the robustness of parameter estimation and the model order. After the application performance cost function is derived, the gradient at the estimated parameter is focused to evaluate the disturbance level of the performance in the application. As a result, in this paper, it is shown that the robustness of parameter estimation is given as 2nd order function of the model order.

View full abstract

In 2001, the Bank of Japan (BOJ) adopted “quantitative monetary easing”. Since short term interest rates became almost zero, the operating target of monetary policy was changed from interest rate to the monetary base, where monetary base is defined as the sum of “Cash” and “Reserve at the BOJ”. Honda et al [1] showed that “Reserve at the BOJ” in (2001,2006) is effective to the economy through a transmission path in a stock market, where impulse responses in VAR model are used in monthly data of Japan. Decomposing money into transaction demand and precautionary one, and estimating precautionary one, Morita and Miyagawa [2] tried to show that increasing “Reserve at the BOJ” makes GDP increased through the stock market in quarterly data. In this paper, the method of estimating precautionary demand in Japan is extensively improved and applied to the case of USA. Using precautionary demand estimated in the whole interval (1980m01, 2012m02), the quantitative easing at Federal Reserve Board(FRB) is shown to be effective during the period (2006m06, 2010m02).

View full abstract

This paper presents a new methodology to compute operational risk using Haar wavelets-based approach, and we illustrate the effectiveness of our algorithms through numerical examples. Additionally, we explain the relationship between the above new methodology and the Bromwich integral.

View full abstract

Integrated Optimization is a new framework of optimization which combines optimization and modeling for higher versatility and practicability. The modeling method that makes a response surface plays an important role for the performance of Integrated Optimization. In this paper, a new modeling method which makes local and global response surfaces is proposed to contribute the global performance in Integrated Optimization. The local response surface is based on locally placing sample points in good solution circumference. The global surface is based on globally placing sample points in the whole feasible area. The effectiveness of the proposed method is examined through numerical simulation using typical benchmark problems.

View full abstract

A stopping rule is developed for finite-difference stochastic approximation (FDSA) which is to minimize an unknown objective function based on random noise corrupted observation of the function. When it is assumed that the function is convex quadratic, the necessary number of iterations for achieving a given probabilistic accuracy of the resultant solution is derived, which gives a rigorous stopping rule for FDSA. This number is polynomial in the problem size.

View full abstract

There are many types of Evolutionary Algorithm. Genetic Programming has the rich expression of the solution for a lot of problems, and Genetic Algorithm is good for the tuning of parameters on array. In our last study, the hybrid algorithm of Genetic Algorithm and Genetic Programming were introduced[1]. We presented the diversified investment using a number of agents obtained with the hybrid algorithm. However, when we apply the presented method, problem of computational time occurs. It has been shown that our model has higher performance to economic time series data rather than others. In this study, we introduce the knowledge data base method, which uses the old solutions as initial solutions for new optimization. In the numerical example, we compare the proposed hybrid algorithm of Genetic Programming including Genetic Algorithm to confirm that our method boosts convergence.

View full abstract

This paper considers a problem on the non-negative derivative constraints on the cubic smoothing spline curves using normalized uniform B-splines as the basis functions. In particular, we derive a condition for monotonic constraints over interval along the line of Fritsch and Carlon's works in [1]. Moreover, we present how these results are incorporated in the optimal smoothing spline problems. The performance is examined by some numerical examples.

View full abstract

An optimization problem with robust and average constraints is introduced, where the constraints are described as parameter-dependent linear matrix inequalities. The difficulty of the parameter dependency is solved by using the scenario approach which employs random samples of the parameter. An explicit number of random samples such that the optimal solution of the scenario problem achieves prescribed accuracy and confidence is derived, which is the main result of this paper. It is then applied to an average pole placement problem with robust stability, and a numerical example is provided.

View full abstract

For describing irregular movement of animals in random search as well as patterns of human travel during daily activity an original model of Brownian nonlinear motion is proposed. The wandering particle is represented as a point in the extended phase space comprising its position, velocity, and, in addition, the acceleration. The acceleration is assumed to be governed by a certain random process with stochastic self-acceleration. The acceleration dynamics is described by a nonlinear stochastic differential equation of the Hänggi-Klimontovich type. Its regular component represents the preference of the particle moving with a certain fixed velocity. The stochastic component with the noise intensity growing with the acceleration is related to the active behavior of the wandering particle in changing the motion direction. The model is studied numerically. The obtained results allow us to state that the developed model generates motion trajectories that can be treated as Lévy random walks. The latter statement can be regarded as the main original point of the present work demonstrating a new approach to modeling the random searching based on continuous Markovian processes.

View full abstract

A Bayesian dynamic method for analysis of software debugging process data is handled. It is adressed to predict remaining errors. In the Bayesian analysis, hierarchical prior models are structured, and empirical and expert knowledge priors are supposed. These priors play roles recognized as representations for complex situations. The empirical prior based on observed data is used for the representation of uncertainty corrections. The prior of success probability for the tests is assumed based on expert knowledge of engineers. The reliability is estimated based on the posterior mean of the number of errors. The Bayesian inferences are derived based on the computational simulation methods, and the information criterion EIC is used to choose appropriate models.

View full abstract

The Brownian motion driven by the chaotic force, which changes chaotically at time interval τ and has the deterministic nature, is discussed in comparison with the conventional Brownian motion, which is driven by the stochastic random noise. The chaotic sequence is generated by some mapping function. In the case of a linear Langevin equation it is shown that for large τ the stationary distribution has the shape of the invariant density of the mapping function and for small τ the stationary distribution can be described by the Gaussian form, which does not depend on the detail of the mapping function. The last case corresponds to the conventional Brownian motion with the stochastic random noise. The above two characteristic stationary distributions are shown to be connected via the fractal structure, if τ is decreased. In the case of a nonlinear Langevin equation the relation between the “chaotic integral” and the stochastic integral is discussed. For small τ the Fokker-Planck equation associated with the nonlinear deterministic Langevin equation is shown to coincide with that of Storatonovich type.

View full abstract

A fractional generalized Cauchy process (FGCP) is studied, which may give the same probability density function as the ordinary generalized Cauchy process. The exact solution of the Fokker-Planck equation for FGCP is given with the aid of the inverse Lévy transform. The associated eigenvalue problem is clarified. It is also exhibited the natures of long-memory, fractal, and volatility clustering associated with the FGCP.

View full abstract

This paper proposes a method invoking one dimensional piecewise continuous map(PCM) in order to investigate the order in chaotic oscillation and periodic oscillation in nonlinear dynamical systems described by ordinary differential equations. The systems studied in this paper are of Van der Pol-Mathieu type. The PCM is derived from the data sequence of state variables of nonlinear systems. Computational experiments on the systems of Van der Pol-Mathieu type and on the PCM show that the proposed method is effective to investigate the existence of chaotic oscillation and periodic oscillation in nonlinear dynamical systems such as Van der Pol-Mathieu type.

View full abstract

In this paper we study the H_∞ tracking problems with preview by state feedback for a class of linear continuous-time systems with impulsive effects and stochastic uncertainties on the finite time interval. The systems include linear stochastic continuous-time systems, linear stochastic discrete-time systems and linear systems with stochastic uncertainties and an input realized through a zero-order hold. The author has already presented the necessary and sufficient conditions for the solvability of these H_∞ tracking problems and given the control strategies for them respectively ([11]). The necessary and sufficient conditions for the solvability of the H_∞ tracking problem are given by Riccati differential equations with impulsive parts and terminal conditions. Correspondingly feedforward compensator introducing future information is given by linear differential equation with impulsive parts and terminal conditions. It is a very important point in this theory. However it has not been yet investigated how the preview feedforward compensator with impulsive parts and effects of stochastic uncertainties of the systems is derived in detail. In this paper we focus on the direct derivation method of noncausal compensator dynamics from the point of view of dynamics constraint. We derive the pair of noncausal compensator dynamics and impulsive Riccati equations by calculating the stochastic first variation of the performance index under the dynamics constraint.

View full abstract

This paper is concerned with the control strategy by vaccination of the infectious disease spread in the populations consisting of the susceptible, the infected and the recovered (SIR). In the realistic spread of the infectious disease, changes in the environment and the weather cause some kinds of random fluctuations in the infection and the recovery rates, etc. Moreover, medical facilities have generally the maximal capacity for treatment of diseases. Taking these facts into consideration, we propose the stochastic infectious model with vaccination and saturated treatment, and we consider the stochastic optimal vaccination problem for the SIR model with saturated treatment using the stochastic maximum principle and the four-step scheme.

View full abstract

In this paper, the optimal design method for the electromagnetic actuator using the electromagnetic field analysis simulator and the metaheuristics optimization is proposed. The proposed method has two routes for a case of heavy simulator load and a case of light simulator load. In the case with a heavy simulator load, Integrated Optimization that combines modeling and the metaheuristics optimization to obtain the solution in practicable time is used. The developed method is applied to the optimal design of the magnetic pole shapes of electromagnetic actuator in the case of a heavy simulator load and the case of a light simulator load, and the usefulness and the practicality of the developed approach is verified.

View full abstract

Human behavior during the process of virtual inverted pendulum balancing in viscous environment is analyzed. The results of the virtual experiments are compared to the results of previous studies on so called dynamical trap effect. It is shown that the phase trajectories and phase variables distributions of the virtual stick motion under human control are similar to those of an oscillator under the presence of noise described by the dynamical trap model. Moreover, it is discovered that the patterns of system dynamics under human control are similar for all feasible values of system parameters. We therefore suggest that the dynamical trap model could reflect certain features of human behavior during processes of dynamical systems control near equlibrium points.

View full abstract

This paper shows a basic approach of stabilizing control and tracking method for keeping a sensing level using a pan-tilt unit on an inverted pendulum mobile robot. The mobile robot has just two wheels, which means that this robot is of an inverted pendulum type. Moreover it has several sensors. In this situation, the pitch angle is not always constant, and the data from the sensors are affected by it. Therefore, it is necessary to obtain precisely their posture such as position and angles, where the sensor devices should be controlled for the sensing level. A framework of tracking control by the pan-tilt unit for the inverted pendulum mobile robot is shown, and further some experimental results are demonstrated on the real robot system KoRo.

View full abstract

This paper presents two approximation schemes for SDEs. The one is a second order scheme simulating stochastic areas, and the other is Heston approximation of stochastic volatility models. For the latter, we give some numerical results.

View full abstract

Coupled human balancing tasks, performed by a pair of artificial controllers and a pair of human subjects, have been studied by the present authors based on the coupled inverted pendula (CIP) model. On the contrary, in this paper, we examine another type of combination, the artificial controller and the human subject. We experimentally estimate Lyapunov and sub-Lyapunov exponents of balancing errors of the system of human subject and artificial controller, in which the human subject is in cooperation with the artificial controller having several different feedback gains. The result implies that the human subject seems to try to make the artificial controller minimally or neutrally stable.

View full abstract

Ionospheric delay is one of the major error sources in GNSS (Global Navigation Satellite System) positioning. For single frequency system, it is very important to effectively correct the ionospheric delay and so-called Klobuchar's ionospheric model has been widely used due to its well known properties such as computational simplicity. The Klobuchar model is one of the global ionospheric models, and it can correct about 50% RMS ionospheric errors[1]. On the other hand, there are some models that can achieve better accuracy than the Klobuchar model, the SCHA (Spherical Cap Harmonics Analysis) model is one of such models. By the SCHA model, it has been reported that the ionospheric delays can be well modeled in the regional area such as the sky over Japan[2]. However, comparing with the Klobuchar model, the SCHA model needs more number of parameters and computational burden. Therefore, in this paper, a method to generate the parameters of the Klobuchar model from the SCHA model is proposed. By this method, the regional ionospheric model that is not only simple and easy operational like the Klobuchar model but also regionally accurate like the SCHA model is obtained.

View full abstract

The number of web cameras installed along shorelines has increased significantly in recent years. Unfortunately, these cameras are overwhelmingly passive in their application: they monitor ocean conditions but do nothing to detect threats and alert other systems. This constitutes a lost opportunity, especially in light of recent hurricane and tsunami disasters. Though such threats can be predicted by meteorological or seismological agencies, localized predictions of the size and arrival times of large waves have not been feasible. Hence, there is a need for a broadly deployable system for detecting wave dynamics. In this paper, we propose such a system. Using standard video footage from web cameras, our system can detect local formation of large waves and thereby reduce their threat potential. Unlike previously proposed wave detection systems based on imaging, such as conventional block matching methods, ours does not require particularly high resolution or stereography, nor is it computationally complex. Based purely on calculating the phase difference between consecutive images of a wave using Fourier transform, our system can detect the displacement of waves with high accuracy.

View full abstract

A residual life prediction system is developed for condition based maintenance of rotating equipment. Faults of the rotating equipment are mainly observed by the abnormal vibration signatures. By monitoring vibration of the equipment, deterioration inclination is estimated and an appropriate maintenance based on the equipment condition can be achieved. The deterioration prediction and the residual life prediction of the rotating equipment can be displayed by using the developed residual life prediction system.

View full abstract

In this study, a hybrid multi-layered Group Method of Data Handling (GMDH)-type neural network algorithm using principal component-regression analysis is proposed and applied to the computer aided image diagnosis (CAD) of liver cancer. In the GMDH-type neural network, heuristic self-organization method, which is a kind of evolutionary computation, is used to organize the neural network architecture. But, multi-colinearity occurs and prediction values become unstable. In this study, hybrid multi-layered GMDH-type neural network using principal component-regression analysis is proposed. In this algorithm, multi-colinearity does not occur and accurate prediction values are obtained. This new algorithm is applied to the medical image diagnosis of liver cancer. First, the GMDH-type neural network which recognizes the liver regions, is automatically organized using multi-detector row CT (MDCT) images of the liver, and the liver regions are recognized and extracted. Then, new another GMDH-type neural network is automatically organized using the extracted image of liver, and the candidate regions of the liver cancer is recognized and extracted. The recognition results are compared with the conventional sigmoid function neural network trained using back propagation method and it is shown that this algorithm is useful for CAD of liver cancer.

View full abstract

By invoking generic restriction on ‘what-to-be-observed', a common information structure is introduced in early perception processes. To identify latent parameters, this structure stabilizes the expectation maximization scheme via dynamic truncation of model errors. Resulted likelihood function can be exploited to control the focus of vision systems.

View full abstract

In this paper, we consider output synchronization of nonlinear agents over switching networks. Communication between the agents is constrained by a switching network. Under the assumption that such a network is identically independently distributed stochastic process, we derive a sufficient condition on feedback controller gains for achieving the output synchronization. In deriving the synchronization condition, the passivity of individual agents plays an important role. A numerical example is included to verify the synchronization condition presented in this paper.

View full abstract

In this study, a hybrid Group Method of Data Handling (GMDH)-type neural network algorithm using principal component-regression analysis is proposed and applied to the nonlinear system identification. The architectures of the GMDH-type neural networks are automatically organized using heuristic self-organization method which is a kind of evolutionary computation.In the heuristic self-organization method, many nonlinear combinations of the input variables are generated and new neurons are constructed from these combinations. Only desirable neurons which fit the characteristics of the nonlinear system, are selected and these neurons are combined again in next layer. These procedures are iterated and a multi-layered neural network architectures are automatically organized. In the GMDH-type neural network, the multi-colinearity of the variables generates and the prediction output values of the neural network become unstable. In this study, the principal component-regression analysis is used for estimating the parameters of the neurons and stable and accurate multi-layered architectures of the GMDH-type neural networks are organized using the heuristic self-organization.

View full abstract

When a neuron in the brain is activated, its membrane potential sharply rises up and immediately drops down. This phenomenon, which is called as a neural spike, is basis of many analyses about the brain in research fields such as neuroscience and neural engineering. Neural spikes are usually recorded by extracellularly voltage recordings and spikes from several neurons are recorded from an electrode. Therefore it has been required to develop the method to detect spikes and to determine which neuron generated each spike. Many previous methods have been proposed, however, they are not robust when more than two spikes are generated closely and their waveforms are overlapped. In this paper, we proposed a method to detect and sort spikes robustly under many overlaps. Extracellularly recorded voltage signals were modeled with hidden Markov model that can generate overlapped spikes. Hidden variables (corresponding to the existence of neural spikes at each time) and model parameters (the shape of spikes from each neuron and the standard deviation of noise) were estimated by alpha-beta algorithm and expectation-maximization algorithm. The number of templates was determined by maximizing BIC. The method was assessed using a simulated signal which contained many overlaps of spikes and additive white Gaussian noise. As the result, it was showed that our method could appropriately detect and sort spikes under many overlaps.

View full abstract

This paper proposes an estimation method for the number of source signals under a two-sensor condition. A joint distribution of observed mixture signals by microphones has the same number of lines as the source signals. Therefore, we propose a number estimation method using Hough transform. The method can estimate using only observed mixture signals. Additionally, we propose a blind source separation method based on the estimated number of the source signals. The proposed methods have been verified by several simulations.

View full abstract

This paper proposes a BSS (Blind Source Separation) method for acoustic signals under a noisy environment. The proposed method can separate the unknown source signals based on the orthogonalization of a joint distribution of observed mixture signals. For a target speech extraction, we use a difference between the joint distribution of speech signals and that of stationary noises. From the simulation results, it is found that the proposed method can separate the original source signals and extract the target speech signal under a noisy environment.

View full abstract

In this paper, we consider phoneme segment discrimination and consonant recognition in order to improve the real-time speech visualization system called ”KanNon”. We show the structure of the system, then we explain the processing of sound signals. The features of the speech signals are obtained by fitting the Auto-Regressive (AR) model to the signals, and phoneme segment discrimination and phoneme recognition are carried out with the features. In addition, we work out experiment of phoneme recognition, and we consider to apply the methods to ”KanNon”.

View full abstract

In a previous paper, we proposed synthetic Brillouin optical time-domain reflectometry (S-BOTDR) to overcome the resolution limit of a conventional BOTDR.We obtained this method's synthetic spectrum by combining the spectrums measured with different composite pump pulses and different composite low-pass filters. In this paper, we solve an optimization problem that minimizes the fluctuation of the synthetic spectrum and experimentally prove that 10-cm spatial resolution, which is much smaller than the conventional BOTDR limit, was attained by the optimal S-BOTDR.

View full abstract