電気学会論文誌C(電子・情報・システム部門誌)
Online ISSN : 1348-8155
Print ISSN : 0385-4221
ISSN-L : 0385-4221
133 巻 , 2 号
選択された号の論文の33件中1~33を表示しています
特集:省電力時代の電子回路技術
巻頭言
特集論文
<電気回路・電子回路>
  • 高橋 康宏, 佐藤 比佐夫
    2013 年 133 巻 2 号 p. 220-225
    発行日: 2013/02/01
    公開日: 2013/02/01
    ジャーナル 認証あり
    This paper presents a low-power sinusoidal supply circuit for our proposed adiabatic logic (2PC2AL). The proposed supply circuit consists of three parts: LC resonant circuit, voltage generator circuit, and peak controll circuit. The proposed circuit is simulated using an 0.18 μm CMOS technology. The simulation results are shown that the power dissipation of proposed voltage generator has 10.8 μW at 100 nA load, and the proposed sinusoidal supply circuit has 35.5 μW at 100 kHz. The power dissipation of the adiabatic/non-adiabatic 16bit MPU decoder are also compared. At transition frequency 10 k to 1 MHz, the proposed decoder with a sinusoidal supply circuit shows a maximum of 75% reduction in power dissipation to that of a static CMOS.
  • 宇都宮 文靖, 吉田 宜史, 道関 隆国
    2013 年 133 巻 2 号 p. 226-233
    発行日: 2013/02/01
    公開日: 2013/02/01
    ジャーナル 認証あり
    A DC-DC converter scheme for thermoelectrically powered wristwatch is described. The converter consists of a variable-stage switched-capacitor (SC) circuit, a clock circuit, and a control circuit. The converter boosts the output voltage of thermoelectric (TE) generator that outputs a voltage of over 0.25V and supplies the generating power to a rechargeable battery with a voltage of 1.3V. A simple maximum-power-point-tracking technique makes it possible to extract power, which is close to maximum power, from the TE generator with a large internal resistance of 1.2kΩ. The maximum power point was automatically obtained by detecting the output open voltage of the variable-stage SC circuit and also by selecting the stage of multi-stage SC circuits. To verify the effectiveness of our converter, we fabricated a prototype with a 0.6μm CMOS process and measured the power extracted from a TE generator. The power extraction efficiency to the converter is 89% and the total conversion efficiency of the power supply module is 29%.
  • 加藤 勇児, 井岡 惠理, 松谷 康之
    2013 年 133 巻 2 号 p. 234-238
    発行日: 2013/02/01
    公開日: 2013/02/01
    ジャーナル 認証あり
    The ΔΣ fractional-N PLL is been researched to realize a low fractional spurious signal characteristic. In this PLL, the ΔΣ modulator sets the fractional division ratio. However, a limit cycle oscillation occurs in the ΔΣ modulator when the input value is fixed. As a result, the limit cycle oscillation increases a spurious signal power. Therefore, some method is required for suppressing this oscillation. In this paper, we propose a self-dithering ΔΣ fractional-N PLL that inhibits the limit cycle oscillation without the external dither generating circuit. The proposed circuit generates the dither from internal signals of PLL. We simulated the output spectrum of the proposed circuit. As a result, we show that the proposed circuit suppressed the limit cycle oscillation, and that the spurious level of the proposed circuit was almost equals to a spurious level without the limit cycle oscillation.
  • 小玉 大地, 井岡 惠理, 松谷 康之
    2013 年 133 巻 2 号 p. 239-244
    発行日: 2013/02/01
    公開日: 2013/02/01
    ジャーナル 認証あり
    The conventional local PWM-DAC for the ΔΣ D/A converter generates the PWM wave by dividing the clock signal. In this method, the sampling speed is decreased greatly by increasing the resolution. In this paper, we propose a new PWM-DAC with the Multi-Delay inverter. This DAC does not require the dividing clock as the conventional PWM-DAC. As simulation results, we show that the 5-th order ΔΣ D/A converter with the propose DAC achieves the S/N of 130dB at the MOS transistor threshold voltage deviation of 100 mV.
  • 和田 和千, 伊藤 遼, 関根 かをり
    2013 年 133 巻 2 号 p. 245-250
    発行日: 2013/02/01
    公開日: 2013/02/01
    ジャーナル 認証あり
    The structure of Proportional-to-Absolute-Temperature (PTAT) voltage reference with MOSFETs operating in the subthreshold region is proposed. It is based on conventional one the equivalent operation of which is being achieved from the viewpoint of relations between voltages and currents of MOSFETs. The derived structure relaxes the occupied area. Experimental results of a PTAT voltage reference implemented on a chip as an example are shown.
  • 矢澤 和樹, 小森山 恵士, 桑原 浩一, 谷本 洋
    2013 年 133 巻 2 号 p. 251-259
    発行日: 2013/02/01
    公開日: 2013/02/01
    ジャーナル 認証あり
    There are strong demmands for low-voltage operating OPA/OTA in deep sub-micron CMOS processes. We have proposed a CMOS inverter based two-stage OTA configuration named F/F+F/B OTA, which can operate 1V to 1.8V of supply voltage, for an 0.18μm CMOS process. In this report, three series of fabrication results for the proposed F/F+F/B OTA configurations are presented with their design details. We have improved common-mode rejection ratio and voltage gain over the prototype chip. The revised chip demonstrated 79dB of differential voltage gain, unity frequency of 3.8MHz, with 124dB of CMRR, from 1.0V of power supply with 200μA of current consumption.
論文
<情報通信工学>
  • 宮崎 保光, 橋本 忠弘, 高橋 港一
    2013 年 133 巻 2 号 p. 260-267
    発行日: 2013/02/01
    公開日: 2013/02/01
    ジャーナル 認証あり
    In recent years, RFID systems have received much attention in information management and security. For evaluation of RFID systems, study of electromagnetic wave propagation and scattering of UHF wave and microwave transmitted from tag antenna to reader antenna in in-door and out-door is indispensable. In this paper, we describe the characteristics of electromagnetic wave scattering, diffraction and interference by obstacles in propagation channel and show the distribution of received level using FDTD method. In RFID systems, receiving characteristics are strongly influenced by the structure of roads and buildings for out-door environment and rooms, doors and windows for in-door environment. Three-dimensional numerical analysis is studied to evaluate the effects of the earth of roads and building with finite heights for out-door applications and effects of ceil and floor of buildings for in-door applications. Signal processing and control are important to recognize each tag code correctly even if multiple codes from tags are received by a reader simultaneously. In this paper, received characteristics when multiple tags transmit signals modulated by ASK are studied by FDTD method and compared with the analytical results.
  • 鈴木 康祐, 旭 健作, 渡邊 晃
    2013 年 133 巻 2 号 p. 268-274
    発行日: 2013/02/01
    公開日: 2013/02/01
    ジャーナル 認証あり
    Recently the light emitting diodes (LED) have been used for lighting and illumination because of their features such as low power consumption and long life. The lighting equipments have only LED. Therefore, it is necessary to install photo detector to enable bidirectional communication. In this paper, we propose a receiving light signal by LED for bidirectional visible light communications. We evaluate the characteristics of LED used as a photo detector for bidirectional visible light communications through experiment. And, we describe the developed experimental system that uses LED as a photo detector. We show that we can use LED as not only light source but also photo detector for bidirectional visible light communications through experiment.
  • 杉垣 文康, 稲井 寛
    2013 年 133 巻 2 号 p. 275-282
    発行日: 2013/02/01
    公開日: 2013/02/01
    ジャーナル 認証あり
    One of the problems in wireless mesh networks includes the HOL (Head of Line) blocking at relay nodes. To overcome that problem, packet scheduling schemes using the MRTS (Multicast RTS) frame, an extension of the RTS frame, have been proposed. The sender first multicasts an MRTS frame to multiple receiver candidates. Then the sender transmits the data packet to the candidate which earliest returns a CTS frame. That mechanism could solve the HOL blocking. In principle, as the number of receiver candidates increases, the transmission successful rate increases, however the overhead also increases. Under the existing scheduling schemes, the sender always arranges more than two candidates. That would increase the overhead. In this paper, to reduce the overhead, we propose a scheduling scheme which restricts the number of candidates is up to two in any time. Via simulation, in comparison with the existing scheme, our scheme provides higher throughput and fairness.
<生体医工学・福祉工学>
  • —有彩色光照明が生体に与える影響—
    野本 洋平, 大無田 孝夫, 大矢 哲也, 澤井 圭, 小山 裕徳, 川澄 正史
    2013 年 133 巻 2 号 p. 283-288
    発行日: 2013/02/01
    公開日: 2013/02/01
    ジャーナル 認証あり
    The popularization of LED and the development of the control technology of the lights provide chromatic lights such as a red, blue, green. The chromatic lights are expected from the psychology and the physiological effect of color. And it is predicted that chromatic lights are innovated positively into a life space. This research is about the image of chromatic lights is examined using heart rate variability and psychological methodology, i.e., paired comparison test. The subjects of paired comparison test were 14 healthy participants (age 21.9±3.7 years old). The subjects of heart rate variability were 7 healthy participants (age 24.0±3.9 years old). We examined from the two viewpoints. 1. In the results of paired comparison test, the score of the white-ice, yellow and green demonstrated greater scores compared with the other. The score of blue, red and pink decreased to lower values compared with the other. 2. In the results of heart rate variability, the mean values of HF of white-Ice, yellow and green demonstrated greater values compared with the other. The mean values of HF of blue, red and pink decreased to lower values compared with the other. It is thus concluded that, using physiological parameters and questionnaire, it is possible to analyze the effect of chromatic lights on living organisms.
  • 林 孝哉, 藤田 悠介, 三谷 芳弘, 浜本 義彦, 瀬川 誠, 寺井 崇二, 坂井田 功
    2013 年 133 巻 2 号 p. 289-296
    発行日: 2013/02/01
    公開日: 2013/02/01
    ジャーナル 認証あり
    Ultrasonic images have been widely used for diagnosis of liver cirrhosis. In practice, there are significant differences in the result of a diagnosis among individuals for physician's experiences. Therefore, the objective and quantitative method for diagnosis is needed. The quality of M-mode image is in general low. Thus, it is very difficult to diagnose liver cirrhosis with M-mode images. In cirrhosis classification using the ultrasonic M-mode image, Zhou's method was proposed. In this paper, compared with the Zhou's method, we propose a method using weighted autocorrelation coefficient to extract an abdominal aorta wall. Furthermore, we propose the use of a bilateral filter to improve the quality of the M-mode images. The bilateral filter is expected to be effective one of smoothing filters preserving an edge to improve image quality in an ultrasonic image. The experimental results show that the weighted autocorrelation coefficient approach leads to improvement in the performance of the abdominal aorta wall extraction and moreover the bilateral filter is effective in both the abdominal aorta wall extraction and cirrhosis classification.
  • 後藤 美穂, 吉田 塁, 斎藤 淳史, 斉藤 亜希, 高山 祐三, 森口 裕之, 小谷 潔, 神保 泰彦
    2013 年 133 巻 2 号 p. 297-302
    発行日: 2013/02/01
    公開日: 2013/02/01
    ジャーナル 認証あり
    To analyze network dynamics, neuronal avalanche is a quite interesting phenomenon. Recently, it was reported that neuronal avalanches were power-law distributed in a slice cortex of a rat, which attracted attention because power-law behavior in neuronal avalanches is an important index representing a network criticality and its efficiency of information transition. In this study, we focused on early stage of network growth and observe avalanche phenomena during their development. Power-law behavior in neuronal avalanches was observed at a matured stage as reported in previous research. However, in early stage, the size distribution of avalanches did not follow power law, indicating that there was the transient period moving on to the critical point.
  • 本間 達, 高木 豊, 若松 秀俊
    2013 年 133 巻 2 号 p. 303-310
    発行日: 2013/02/01
    公開日: 2013/02/01
    ジャーナル 認証あり
    Brain hypothermia requires controlling its temperature within an appropriate range, considering the change of body temperature in a long period. Various mathematical models have been used for the study of control and cooling capability of brain temperature in hypothermia. In the previous models, a hemisphere in a lumped parameter of a uniform temperature has been assumed as a simplified brain without considering the temperature distribution. In the present study, however, a new model is proposed to visualize the temperature distribution in the brain. The model has an approximate shape of each organ in a head based on MRI data, and may well reflect the properties such as heat transfer coefficients, metabolic heat production and heat capacity of human organs. The model has a pseudo-blood-flow model in which any temperature can be set as initial value at the starting place of blood flow. Some simulations using this model are performed with its controlled temperature by the introduction of Ringer's solution into any of the four arteries to the brain. The results of simulation suggests that the various cooling effects are made clear in every region of brain, and that the temperature distribution can be known for the application of controlling brain temperature in a concerning part.
<システム・計測・制御>
  • 西 恵理, 若松 勇貴, 武井 安津子, 新川 拓也
    2013 年 133 巻 2 号 p. 311-315
    発行日: 2013/02/01
    公開日: 2013/02/01
    ジャーナル 認証あり
    The aim of this study was to clarify the mechanical action of the infant tongue during sucking. We developed an artificial nipple containing multiple small built-in force sensors capable of measuring forces a tongue applies to an artificial nipple. A force sensor in the shape of a cantilever structure composed of a thin stainless steel beam and all-purpose foil strain gauges were attached to the surface of the beam. The signals obtained from the force sensor underwent amplification, and subsequently were transferred to a PC via a USB port. Using this system, measurements were taken from three infant test subjects. Institutional approval had previously been granted. Waveforms indicating the elevation of the tongue movement from the apical to the posterior part of the tongue, and the main tongue force applied from under the nipple, were observed.
  • 南角 茂樹, 水篠 公範, 小泉 寿男, 福田 晃
    2013 年 133 巻 2 号 p. 316-325
    発行日: 2013/02/01
    公開日: 2013/02/01
    ジャーナル 認証あり
    In a system with a real-time operating system (RTOS), it is general to use a task which realizes the concurrence or to use a semaphore which realizes a mutual exclusion. But about 30% of systems do not employ OS. In the systems without RTOSs, it is general to employ the Interrupt Service Routine (ISR) for realizing the concurrence and to use Interrupt Enable/Disable for realizing the mutual exclusion among several ISRs. Unlike the Semaphore RTOS provides, however, if a system uses the Interrupt Enable/Disable, there is a problem that a speed of the system becomes slow. This is because other ISRs which have not relation with the ISR cannot be executed by the Interrupt Disable mechanism. In addition, for real-time systems, a processing related with ISRs of higher priority may be not finished in a given time limit due to the above reason. Using the proposed method that adopts semaphores to achieve exclusive control among ISRs, the real-time multi-processing of ISRs was achieved successfully in the embedded systems without an RTOS. Therefore, in the systems without RTOS, we have proposed and implemented Real-Time Embedded Monitor called REMON. With REMON, the mutual exclusion which influences only concerned ISRs can be realized.
  • 尾形 幹人, 薦田 憲久
    2013 年 133 巻 2 号 p. 326-333
    発行日: 2013/02/01
    公開日: 2013/02/01
    ジャーナル 認証あり
    Deduplication backup technology removes redundant data segments over the system to reduce the capacity in target backup storage. This technique also provides better performance, less resource utilization, less energy consumption and TCO. This paper describes an optimization method for deduplication backup in IT system in which multiple duduplication processes are simultaneously installed and activated. The method provides the assignment algorithm of backup target files to installed deduplication processes to maximize aggregate deduplication ratio under keeping the predefined system requirements, such as Backup-windows limitation, Resource utilization limitation. In the practical system, the consuming time to perform deduplication in each process is not constant, but variable caused by several operational resource contention, waiting time in queues, characteristics of data and so on.
    The proposed method point out these time parameters can be simulated to follow a normal distribution, formularize the discrete assignment programming, then define the combinatory approaches of integer linear programming to incorporate the maximal deduplication ratio and the binary adjustment of parameters to ensure being within the variance. By applying the method, system can achieve the maximal deduplication ratio under keeping the time requirements with predefined tolerance. Effectiveness of the method is proven by simulation.
  • Hugang Han
    2013 年 133 巻 2 号 p. 334-341
    発行日: 2013/02/01
    公開日: 2013/02/01
    ジャーナル 認証あり
    An approach to adaptive fuzzy sliding-mode control (SMC) for a class of nonlinear dynamical systems with lumped parameter uncertainties accompanied with control input is proposed and analyzed. In particular, the approach is able to deal with the T-S fuzzy model with different control matrices in the fuzzy rules. The approach guarantees that the system state will be uniformly ultimately bounded. Also, discussion on how to alleviate the conservativeness of the approach is provided. Finally, computer simulations illustrate the effectiveness of the control approach proposed in this paper.
  • 戸石 大輔, 小中 英嗣
    2013 年 133 巻 2 号 p. 342-349
    発行日: 2013/02/01
    公開日: 2013/02/01
    ジャーナル 認証あり
    The configuration of a two-wheeled vehicle, such as Segway, cannot be stabilized by continuous and time-invariant state feedback due to its non-holonomic constraints. Because of the nonlinear nature of the nonholonomic constraints, the realization of a model predictive control (MPC) for this class of vehicles is a difficult task.
    This paper proposes a MPC method that can achieve long prediction horizon and quick computation. At the first step, the optimization of an input (i.e., velocity and steering) sequence is formulated as a graph search problem by restricting the inputs to discrete values. Next, in the second step, the optimized control result is learned by machine learning method, such as SVM.
    A longer horizon MPC compared to that with nonlinear optimization can be realized. The advantages of the proposed method are demonstrated with simulation and experimental results.
  • 武田 明, 髙田 和之, 長尾 尋智, 王 建青, 藤原 修
    2013 年 133 巻 2 号 p. 350-355
    発行日: 2013/02/01
    公開日: 2013/02/01
    ジャーナル 認証あり
    For evaluating the quality of hemodialysis from the limited volume of human blood using a commercially available open-ended coaxial probe, we previously measured for 10 normal healthy subjects and 9 patients who require hemodialysis the dielectric properties of 2.5ml whole blood with a temperature of 25 degrees centigrade in a syringe with a diameter of 20mm and a depth of 8mm. In this study, we measured the complex relative permittivity of water and human blood when changing temperature from 25 to 37 degrees centigrade in order to examine the correlation between them, and calculated their Pearson product-moment correlation coefficients for εr′ and εr′′ along with significant probability P based on a t-test. It should be noted that P is a reference probability to determine whether or not a null hypothesis can be rejected, and that the P value of 0.05 is commonly used as a significance level for a statistical test. As a result, it was found for the complex relative permittivity of water that there is approximately good agreement between the measured results and calculated ones from the Kaatze formula expressed as a function of temperature, and that strong correlations with P < 0.05 are also confirmed between the complex relative permittivity and temperature. For the human blood, on the other hand, we found that a strong correlation with P < 0.05 is observed between εr′′ and temperature, while there is not always significant correlation with P > 0.05 between εr′ and temperature.
  • 萩原 良信, 庄司 達矢, 今村 弘樹
    2013 年 133 巻 2 号 p. 356-364
    発行日: 2013/02/01
    公開日: 2013/02/01
    ジャーナル 認証あり
    In this research, position and rotation estimation for mobile robots in outside of a recording path is realized by applying ego-motion to view-based navigation. The ego-motion is calculated from differences in 3D-positions of SURF feature-points between recording and current images obtained by a Kinect sensor. In conventional view-based navigations, it is difficult to plan another path when humans and objects are located on the recording path. By using our proposed estimation, it is possible to realize flexible path planning in actual environments which include humans and objects. From experimental results performed in actual indoor environments, we evaluated mesurement accuracy for position and rotation estimated by our method, and confirmed the possibility of our method for actual environments including humans and objects.
  • 武藤 康彦, 原 智啓
    2013 年 133 巻 2 号 p. 365-372
    発行日: 2013/02/01
    公開日: 2013/02/01
    ジャーナル 認証あり
    The linear time-varying multivariable system whose controllability indices or observability indices are not constant is called the non-lexicographically-fixed system. Valasek proposed the pole placement design method for such a continuous system by augmenting the system equation so that the augmented system is lexicographically-fixed. This paper concerns the pole placement and the observer design method for linear time-varying discrete non-lexicographically-fixed system. Using the Valasek's idea, the procedure to augment a discrete non-lexicographically-fixed system to a lexicographically-fixed augmented system will be presented. Then, the simple pole placement technique can be applied to the augmented system without transforming the system into any canonical form. Further, using the property of the anti-causal dual system, it will be shown that the pole placement design method can be used for the augmented observer for non-lexicographically-fixed systems. Finally, the stability and the separation principle of the total closed loop system are also shown for the case where both of the augmented pole placement controller and the augmented observer are used.
<メディア情報,ユーザ・インタフェース>
  • 松本 翔平, 綿谷 亮, 岩井 大輔, 佐藤 宏介
    2013 年 133 巻 2 号 p. 373-379
    発行日: 2013/02/01
    公開日: 2013/02/01
    ジャーナル 認証あり
    This paper presents a writing state classification technique by a sedentary behavior sensing with a sensor equipped chair, which wirelessly monitors user's states such as his/her behaviors or physiological and psychological conditions, for the estimation of the user's subjective difficulty of a studying problem on a desktop. Avoiding user's discomfort, four load cells are equipped behind the seat plate of a regular office chair. The system simply measures the weight and the CoP (Center of Pressure) of the seat. These data are divided into time segments, which are labeled into four sedentary body sway primitives by the decision tree algorithm, and then the behaviors are derived from the number and the pitch of those primitives in a certain interval. System evaluation experiments resulted that it achieved more than 80% of labeling accuracy for the untrained users. We finally conducted a user experiment that shows its potential for desk work applications. The result of the experiment showed that our proposed system could classify user's states into writing state and the other states such as reading a document and watching movie at the classification rate of 87%.
<音声画像処理・認識>
  • 小林 洋介, 近藤 和弘
    2013 年 133 巻 2 号 p. 380-387
    発行日: 2013/02/01
    公開日: 2013/02/01
    ジャーナル 認証あり
    We investigated the clustering of non-stationary noise used in the subjective and objective assessment of speech intelligibility. The feature vector used in the clustering comprises 15-dimensional features used typically in MIR and clustering was done using the x-means method. We then conducted tests to validate the clustering results using the Japanese Diagnostic Rhyme Test. As a result, with the JEIDA-NOISE database, the noise can be classified into three clusters, and significant difference between speech intelligibility of each cluster was seen. Finally, we tested the objective speech intelligibility assessment for each cluster using the fwSNRseg and the logistic function. As a result, the performance of objective assessment improved by about 0.01 compared to the case without clustering.
<ソフトコンピューティング・学習>
  • Chuan Yue, Shingo Mabu, Kotaro Hirasawa
    2013 年 133 巻 2 号 p. 388-397
    発行日: 2013/02/01
    公開日: 2013/02/01
    ジャーナル 認証あり
    Along with the explosive development of electronic commerce, trading goods online becomes much more popular and the trading volume over internet has been increased hugely. Concentrating particularly on continuous double auction (CDA), which is an efficient market mechanism, this paper studied and discussed a Genetic Network programming (GNP) based bidding strategy with adjusting parameters for autonomous software agents in agent-based large-scale CDAs (GNP-AP). GNP is one of the evolutionary computations, and the individuals with directed graph structures, represents the potential bidding strategies. Combined with the heuristic control rules, each individual can collect and judge the auction information, then choose the decision-making transition depending on the judgment results. The parameters for helping to select the right decision are adjusted during the evolution in order to get more profits for large-scale CDAs. In the experiments, we studied and discussed the performance of the proposed bidding strategies, and compared it with other classic bidding strategies and previous GNP-RN strategy in large-scale CDA under different settings.
<情報処理・ソフトウェア>
  • 渡辺 透, 小泉 寿男
    2013 年 133 巻 2 号 p. 398-409
    発行日: 2013/02/01
    公開日: 2013/02/01
    ジャーナル 認証あり
    This paper proposes an M2M client/server system with Computer Telephony Integration (CTI) that exchanges information between a human and objects in a Machine to Machine. In this method, HTTP protocol is used for the PULL-type communication and SIP protocol is used for the PUSH-type communication to perform bi-directional communication between the M2M client and the M2M server. Accordingly, a CTI server having a SIP server function and a Web server function is used for the M2M server. Furthermore, a Smart Phone having a SIP phone function as well as a Web browser function is used for the M2M client. In this study, an M2M system of a M2M client/server system with CTI functions is implemented and evaluated.
  • 岩井 亮, 加藤 昇平
    2013 年 133 巻 2 号 p. 410-417
    発行日: 2013/02/01
    公開日: 2013/02/01
    ジャーナル 認証あり
    Differential Evolution (DE) is a powerful stochastic algorithm for real-coded optimization. However, DE has a problem as well as other traditional stochastic optimization algorithms: that it is difficult to optimize areas that are little globally convex. Thus, it is difficult for DE and traditional algorithms to optimize some practical problem where globally convex cannot be supposed. To solve this problem, we propose Differential Evolution on Scattered Parents (DE-SP) that re-selects the individuals on each dimension when the mutant individual is calculated and some children individuals' candidates unconditionally become the children individuals. We have implemented three types of optimization experiment to verify the effectivity of DE-SP: Noisy Function 1 (NF1), that is a benchmark problem with little globally convex, Noisy Function 2 (NF2), that is the one with globally convex, and an optimization problem for bipedal robot to stand stably. Thereby, we confirmed that DE-SP was the most stable algorithm to optimize areas that is little globally convex among the comparative existing algorithms: DE, DE/nrand/1, DE/isolated/1, Hybridizing Particle Swarm Optimization with Differential Evolution, and Wavelet-Mutation-Wavelet-Crossover-Based Differential Evolution, and found the best objective function value at the practical problem.
  • 大野 誠寛, 村田 匡輝, 松原 茂樹
    2013 年 133 巻 2 号 p. 418-426
    発行日: 2013/02/01
    公開日: 2013/02/01
    ジャーナル 認証あり
    To generate readable captions for Japanese spoken monologue such as lectures in real time, it is necessary to sequentially display the caption into which proper linefeeds are inserted. This paper proposes a technique for sequentially inserting proper linefeeds into the lecture transcript whenever a bunsetsu, which is a linguistic unit shorter than a sentence in Japanese that roughly corresponds to a basic phrase in English, is identified. Assuming that linefeeds are inserted into bunsetsu boundaries, the technique can make the delay time of captioning shortest. The technique statistically judges whether or not a linefeed should be inserted into each bunsetsu boundary by using the information which is available at the time. We conducted experiments on linefeed insertion using a Japanese lecture corpus. From the experimental results, we confirmed that our method, which is bunsetsu-based linefeed insertion method, had almost as much accuracy as the sentence-based linefeed insertion method. In addition, we conducted the comparative evaluations with four baseline methods. As the results, we confirmed that our method could insert linefeeds more properly than the simple methods which are thought to have as same delay time as our method.
研究開発レター
<情報通信工学>
<情報処理・ソフトウェア>
 
部門記事
 
feedback
Top