In this study, two coupled chaotic oscillators composed of RC circuits are investigated. We carry out computer simulations and circuit experiments and observe both in-phase synchronization and anti-phase synchronization. Moreover, when we change the coupling strength between oscillators, we observe a change in synchronization state.

View full abstract

In this study, we investigate clustering patterns generated in coupled chaotic circuits in networks. In these networks, the coupling strength reflects the distance information, and each chaotic circuit is connected to all other chaotic circuits. We consider the relationship between coupling strength and phase difference by changing the scaling parameter of coupling strength. Furthermore, we determine the various phase synchronization patterns when we change the number of chaotic circuits.

View full abstract

The conventional hardware/software codesign cannot meet the requirement for a PSO algorithm that needs to be changed for different applications. In addition, the architecture of the conventional hardware/software codesign leads to a low processing speed. This paper presents a flexible hardware/software codesign to support various PSO algorithms and increase the processing speed of the conventional hardware/software codesign for various PSO applications. By adopting a simplified hardware architecture, the calculation speed can be improved by reducing the communication overhead. Furthermore, an improved generic particle calculation block (GPCB) is used to enhance the flexibility. It selects an appropriate PSO algorithm without the need for hardware modification to further increase the processing speed. The experimental results proved that the proposed flexible hardware/software codesign can achieve high processing speed and high flexibility with low chip cost.

View full abstract

Recently, a novel algorithm of filter-based single-image super-resolution (SR) has been proposed. We here propose a hardware-oriented image-enlargement algorithm for the SR algorithm based on frame-bufferless box filtering, and present novel circuits of the proposed enlargement algorithm and the SR algorithm for an field-programmable gate array (FPGA), aiming at the development of single-image SR module for practical embedded systems.

View full abstract

In recent years, a bike-sharing system (BSS) has been introduced as a means of inner city transportation in cities of various countries. The BSS consists of one control center and many dispersed stations where rental bicycles are parked. After many people use the bicycles, the distribution of bicycles among stations changes from the managed initial distribution to a disarranged one. Hence, it is necessary to restore the disarranged distribution to the initial one at the certain time intervals to sustain the good performance of the BSS. This work is done by vehicles that transport bicycles from stations with excess bicycles to ones with a shortage. How to accomplish this task as efficiently as possible is an interesting issue. We have presented a mathematical optimization model on this issue named the BSS routing problem (BSSRP) to determine the shortest tour of the transport vehicle, and proposed heuris-tic solution methods to solve it. In this paper, we propose two new solution methods for the BSSRP. One is based on the idea of searching for solutions including the infeasible space, and the other is based on the idea of dividing the search process into two phases. The results of numerical experiments show that both newly proposed methods are superior to previous methods.

View full abstract

We analyze the effectiveness of a multi-armed bandit algorithm, which utilizes ideal spatiotemporal chaotic dynamics generated by an FIR filter. In the previous research on additive chaotic noise to heuristic searches for combinatorial optimization problems, it has been shown that the chaotic sequences with negative autocorrelation improve the performance of asynchronously updated algorithms. The effectiveness of chaos can be understood in terms of the conventional theory of the chaotic CDMA, which showed that the cross-correlation between the sequences with negative autocorrelation becomes lowest. The spatiotemporal chaotic searching dynamics with such lowest cross-correlation has been shown to be effective in improving asynchronously updated algorithms. In this paper, as an asynchronously updated multiarmed bandit algorithm, we apply the FIR filter to the softmax algorithm, and analyze the spatiotemporal dynamics of this method. Our numerical simulation results show that the cross-correlation of this method can be minimized and its performance can be improved by using negative autocorrelation.

View full abstract

In future wireless and/or mobile networks, voice, highcapacity data, and multimedia contents might be converged onto a single wireless network platform, which leads to increased complexity and heterogeneity. To realize such advanced communications, it is necessary to consider users' QoS requirements. In our related work, we proposed a novel QoS framework based on cross-layer design that is jointly optimized among all layers. In this paper, we propose a detailed protocol scheme, that is, a management scheme of users' QoS requirements for wireless heterogeneous network. Moreover, we reveal the effectiveness of our framework.

View full abstract

In downlink multiuser multiple-input multiple-output (MIMO) orthogonal frequency division multiplexing (OFDM) systems, a base station (BS) communicates with multiple user terminals (UTs) simultaneously using MIMO transmission. Since UTs each receive signals intended for other UTs in addition to those intended for themselves, they suffer from multiuser interference (MUI). To eliminate MUI, the use of block diagonalization (BD) has been proposed. BD is a scheme of channel inversion that divides a multiuser MIMO channel into multiple single-user MIMO channels by transmitter-side linear precoding. However, BD has two restrictions about the antenna assignment to be able to calculate the precoding weight matrix. In this paper, we propose another linear precoding scheme for MUI elimination. Our proposed scheme is released from one of two restrictions in calculating the precoding weight matrix for MUI elimination. Simulation results show that the proposed scheme and BD achieve the same bit error rate (BER) performance under the same antenna assignment. Furthermore, we can show that the proposed precoding scheme eliminates MUI even under antenna assignments which are impracticable with BD and that the principle of the proposed scheme is applied to uplink multiuser MIMO-OFDM systems.

View full abstract

Artificial Bandwidth Extension (ABE) is a technique that attempts to regenerate the wideband signal (0-8 kHz) from a narrowband signal (300-3.4 kHz) in order to improve the speech quality of today's telephone systems. By employing the well-known source-filter model, ABE can be divided into 2 sub-problems, namely extension of the excitation signal and extension of the spectral envelope. Since the extension of spectral envelope has a more significant effect on perceived speech quality as compared to that of the excitation signal [1], more efforts have been put into this problem.
Several approaches to the problem of extending the spectral envelope have already been proposed. Although speech quality is improved as compared to the narrowband speech, a clear gap between the regenerated speech and original wideband speech could still be seen. In this paper, we propose a new approach to this problem by introducing a discriminative model into SPLICE [2], and describe its performance.

View full abstract

To meet classic but still intrinsic requirements of speech synthesis in tonal languages, the tone nucleus model is employed in Thai. Tone nuclei are defined for all five distinctive tones according to the underlying targets. The full process of F₀ contour generation is presented from the nucleus extraction until the F₀ contour generation for the continuous speech. The objective and subjective tests confirm the efficiency and adaptability of the model in the other language than Mandarin. Compared to the F₀ contours generated by the predictors trained from the contours in the whole syllables without extracting the tone nuclei, the model generates the F₀ contours in the continuous utterances with less distortion from the target but more tone intelligibility and more naturalness. The modified method is also introduced to enhance the accuracy in the F₀ generation by the model.

View full abstract

We describe a new method of voice conversion aimed at character conversion by the eigenvoice Gaussian mixture model (EV-GMM) approach. Using an eigenvoice space built from 273 speakers and speech samples of three different characters created by a single skilled voice actor/actress, the conversion can generate the voices of the three characters from an arbitrary speaker, while retaining the speaker identity. Listening tests were carried out by presenting two kinds of synthetic voices: those before and after the character conversion. The results showed that listeners, both native and non-native speakers, can perceive well the character voice difference intended by experimenters. Moreover, our proposed method performs better than the F₀-based approach. The understanding of how to create synthetic speeches that realize character conversion within the same individual reveals a necessity for and chances to develop a more intelligent correct feedback for voice training as well as a voice therapy system, which will provide participants with an idea of what their expected voices should be.

View full abstract

In this paper, we aim to construct an evaluation method for bicycle-riding comfort using biological signals. In recent years, it has become important, in bicycle development companies (e.g., MIYATA CYCLE Co., Ltd.), to develop products by reflecting human sensibility in order to improve user satisfaction. In addition to conventional evaluation methods using acceleration, angular velocity, and questionnaires, we focus on the electromyogram (EMG) and the electroencephalogram (EEG) as new evaluation indexes. We analyze the EMG and the EEG to estimate the features of these biological signals and how they reflect changes in bicycle-riding comfort.

View full abstract

We propose a new alignment algorithm for low-cost performance image plane tracking. The proposed method consists of fast 2D-based tracking and accurate 3D-based alignment steps. In order to confirm the effectiveness of the proposed method, we demonstrate the simulation experiments and show that our method improves the computing cost of the conventional method.

View full abstract

In this paper, we propose a sound-quality-evaluation system using an electroencephalogram (EEG) and a group method of data handling (GMDH) type neural network. EEGs are used in various applications, and we focus on sound quality evaluation. We prepared EEG samples to train a GMDH-type neural network to recognize 3 typical types of sound that were used to create the training data. The results showed that using the GMDH-type neural network improved the recognition rate compared with other methods.

View full abstract

The growing number of surveillance cameras installed for crime prevention in various city locations capture images that might reveal private information, such as images of human faces, of car license numbers, and of windows, without permission. We previously reported a Hierarchical Masked image Filtering (HMF) technique that can mask and restore such objects in a captured image [1]. The object detection used for masking the privacy-related images uses the Haar-like [2] feature algorithm in the OpenCV library. We have now improved the HMF masking performance by developing a fast object detection method that is used instead of the Haar-like algorithm.
In this paper, we describe HMF based on the morphological pattern spectrum. This pattern spectrum can be used to mask not only a front face image but also a profile image. The use of this HMF reduced the processing time by about 62% compared with that of the Haar-like algorithm.

View full abstract

We propose a novel method for recognizing human actions from video sequences using multiclass support vector machines (SVMs). The proposed method enables the optimal combination of different space-and-time features. Our method is categorized as a deep model in which SVMs at the first-level are used to generate class probabilities based on each frame in the videos. At the second level, the SVM finally classifies actions by combining class probabilities of each video stream through the action time interval. Our twolevel SVM architecture can also combine different kinds of features efficiently and give better results than Bayes combination. We used two kinds of sparse features, i.e., optical flows and ST patches. Our experiments using the KTH dataset show that the two-level SVM architecture is effective for combining different kinds of features through action time intervals.

View full abstract

For solving inverse problems in computed tomography (CT), we have presented a continuous-time image reconstruction (CIR) system described by a switched nonlinear dynamical system with a piecewise smooth vector field. Recently, we have also proposed an extended CIR (ECIR) system with a modification of the nonlinear vector field, for the purpose of solving inverse problems in binary tomography, which is the process of reconstructing a binary image from a finite number of projections. In this study, we show, by means of numerical experiments, that the ECIR system is effective not only for binary tomography, but also for continuous tomography, such that the maximum pixel value to be reconstructed is previously given. For projection data with insufficient views, we found that the ECIR system exhibited a better convergence performance than the maximum-likelihood expectation-maximization method.

View full abstract