Improvement in learning efficiency of multilayered neural networks based on dynamic systems

This paper proposes a new method which uses the dynamic systems for improving learning efficiency in multilayered neural networks. The shape of the error function is analyzed using Monte Carlo simulation, and the learning efficiency is improved on the basis of the shape analysis and the adaptive learning rate. The proposed method uses the adaptive learning rate which is determined from the gradient of the error function. The proposed method is applied to XOR and 5-bit parity check problems and the effectiveness and feasibility of the proposed method are verified. © 1999 Scripta Technica, Electr Eng Jpn, 129(4): 77–86, 1999     

Analysis of the electric field and force in an arrangement of a conducting sphere and a plane electrode with a dielectric barrier

This paper presents an analysis of the electric field and dielectrophoretic force in an arrangement consisting of an uncharged conducting sphere and a plane electrode with a dielectric barrier. The electric field is calculated by using the method of multipole images using an iterative algorithm proposed for calculating the images of the dielectric barrier of finite thickness. The calculation results show electric field intensification due to the presence of the dielectric barrier having higher permittivity, /spl epsiv//sub S/, than that of the surrounding medium, /spl epsiv//sub E/; however, if the barrier is separated from the conducting sphere by at least the sphere radius, its influence is negligible. Inside the dielectric barrier, the electric field on the axis of symmetry becomes more uniform and the average field significantly increases with decreasing its thickness. For a case where dielectric barrier is sufficiently thin, the electric field at the contact point and the force on the conducting sphere vary approximately as power functions of /spl epsiv//sub S///spl epsiv//sub E/.     

First plasma experiment on spherical tokamak device UTST

The UTST (University of Tokyo Spherical Tokamak) device was constructed for the purpose of exploring the formation of ultrahigh‐beta ST (spherical tokamak) plasma using the double null plasma merging method. When two plasmas merge together to form a single plasma, magnetic field lines reconnect, and magnetic field energy is converted to plasma kinetic energy, increasing the plasma beta. Merging start‐up has been demonstrated in the TS‐3/4, START, and MAST devices using coils inside the vacuum vessel, and the TS‐3 plasma obtained a 50% beta. In order to demonstrate start‐up in a more reactor‐relevant situation, UTST has all poloidal field (PF) coils outside the vacuum vessel. The first plasma experiment on the UTST was begun in December 2007. In the results, the plasma obtained 10 kA by using only the outer PF coils and a single ST was generated in the lower area (z = –0.3 to –1.0 m) close to a washer gun. This result suggests that another washer gun on the top of the UTST can allow the generation of ST in the upper area and merging start‐up by using outer PF coils. © 2012 Wiley Periodicals, Inc. Electr Eng Jpn, 179(2): 20–26, 2012; Published online in Wiley Online Library ( wileyonlinelibrary.com ). DOI 10.1002/eej.21216     

Evaluation of the reduction in CO2 emissions by applying Micro‐Grid to home energy supply system

Dispersed generators such as wind power systems, photovoltaic systems, and cogeneration systems are expected to mitigate the environmental burden of energy consumption, and their installation has been promoted recently. Micro‐Grid is focused on as a method to solve some problems in a commercial electric power line when installing a large number of dispersed generators, and some demonstrative research on Micro‐Grid for large‐scale systems is being carried out now. Also, small cogeneration systems for houses, such as gas engines and fuel cells, are expected to improve CO₂ emissions. However, if the power and heat demand of a family are relatively small or are unbalanced, the cogeneration system does not operate effectively. The authors have studied the application of Micro‐Grid for home energy supply, and have developed a control system to solve this problem. The system achieves a reduction of CO₂ emissions and energy costs by sharing electric power and heat among some houses with cogeneration systems. This paper presents an outline of the newly developed system, and in particular describes the effect of the reduction in CO₂ emissions compared with a conventional energy supply method, and the case in which dispersed generators are installed in some houses and operate independently. © 2009 Wiley Periodicals, Inc. Electr Eng Jpn, 170(3): 19–27, 2010; Published online in Wiley InterScience ( www.interscience.wiley.com ). DOI 10.1002/eej.20879     

Involvement of Cytokines in the Pathogenesis of Diabetic Macular Edema

Diabetic macular edema (DME) is a critical complication of diabetic retinopathy, a condition that arises from the breakdown of the blood–retinal barrier and the consequent increase in vascular permeability. Over the years, attempts have been made to treat DME by various approaches, including laser photocoagulation, steroid triamcinolone acetonide, and vitrectomy. However, treatment was unsatisfactory until research identified vascular endothelial growth factor (VEGF) as a factor in the pathogenesis of DME. Intraocular anti-VEGF agents show good efficacy in DME. Nevertheless, in some patients the condition recurs or becomes resistant to treatment, suggesting that other factors may be involved. Because inflammation and retinal hypoxia are seen in DME, research has examined the potential role of cytokines and other inflammatory mediators. In this review, we provide an overview of this research and describe feedback mechanisms that may represent a target for novel treatments.     

Low Frequency Susceptibility of High-Density bcc Solid 3He in the Paramagnetic and Nuclear-Ordered States

We have measured the AC magnetic susceptibility and static magnetization of high-density bcc solid ³ He through the nuclear-ordering transition. The susceptibility in the paramagnetic state strongly depends on the frequency of the measuring field. Near the transition temperature a sharp peak in the real part of the AC susceptibility and an abrupt depression in the imaginary part are observed. The transition temperature indicated by the AC susceptibility is higher than that derived from static magnetization. We analyzed the new behavior in the susceptibility in terms of the spin relaxation between the Zeeman system and the exchange system. The relaxation time in the energy flow in the two systems is in the range of milliseconds in the paramagnetic state, and decreases drastically by two orders of magnitude in the ordered state. The relaxation time in the paramagnetic state is interpreted as due to exchange narrowing, while in the ordered state is explained to be the drift time of the spin wave limited by the size of the sample grown in the pores of the sintered silver. The ordering temperature is given as a function of molar volume in the entire range of the bcc phase.     

Metastable precipitation in isothermally solidified YAG-alumina composites with off-eutectic composition

YAG-alumina composites with Al₂O₃-rich and YAG-rich compositions were prepared by annealing the hot-pressed composites at the eutectic temperature (1820°C) for 1 h followed by cooling down at 10 or 100°C/min in argon atmosphere. Microstructural observations of the solidified composites identified that metastable phase -Al₂O₃ precipitated in the Al₂O₃-rich specimens cooled down at both the cooling rates. For the solidified YAG-rich specimens, however, equilibrium phase -Al₂O₃ could not be reprecipitated, moreover, metastable YAlO₃ phase was formed at cooling rate of 100°C/min. Formation of the metastable phases in the solidified composites was attributed to the decreased melting temperature caused by the divorced eutectic precipitation and the metastable decomposition of YAG component in YAG-alumina eutectic melt during rapid solidification.     

Platform design for large-scale artificial market simulation and preliminary evaluation on the K computer

Artificial market simulations have the potential to be a strong tool for studying rapid and large market fluctuations and designing financial regulations. High-frequency traders, that exchange multiple assets simultaneously within a millisecond, are said to be a cause of rapid and large market fluctuations. For such a large-scale problem, this paper proposes a software or computing platform for large-scale and high-frequency artificial market simulations (Plham: /pl\(\Lambda\)m). The computing platform, Plham, enables modeling financial markets composed of various brands of assets and a large number of agents trading on a short timescale. The design feature of Plham is the separation of artificial market models (simulation models) from their execution (execution models). This allows users to define their simulation models without parallel computing expertise and to choose one of the execution models they need. This computing platform provides a prototype execution model for parallel simulations, which exploits the variety in trading frequency among traders, that is, the fact that some traders do not require up-to-date information of markets changing in millisecond order. We evaluated a prototype implementation on the K computer using up to 256 computing nodes.     

Recurrent infomax generates cell assemblies, neuronal avalanches, and simple cell-like selectivity

Recently multineuronal recording has allowed us to observe patterned firings, synchronization, oscillation, and global state transitions in the recurrent networks of central nervous systems. We propose a learning algorithm based on the process of information maximization in a recurrent network, which we call recurrent infomax (RI). RI maximizes information retention and thereby minimizes information loss through time in a network. We find that feeding in external inputs consisting of information obtained from photographs of natural scenes into an RI-based model of a recurrent network results in the appearance of Gabor-like selectivity quite similar to that existing in simple cells of the primary visual cortex. We find that without external input, this network exhibits cell assembly-like and synfire chain-like spontaneous activity as well as a critical neuronal avalanche. In addition, we find that RI embeds externally input temporal firing patterns to the network so that it spontaneously reproduces these patterns after learning. RI provides a simple framework to explain a wide range of phenomena observed in in vivo and in vitro neuronal networks, and it will provide a novel understanding of experimental results for multineuronal activity and plasticity from an information-theoretic point of view.     

4032 ppi High‐resolution OLED microdisplay

A 0.5‐inch Ultra Extended Graphics Array (UXGA) organic light‐emitting diodes microdisplay has been developed with 6.3 μm pixel pitch. Not only 4032 ppi high resolution but high frame rate, low power consumption, wide viewing angle, and high luminance have been achieved. This newly developed organic light‐emitting diodes microdisplay is suitable for Near‐to‐Eye display applications, especially electronic viewfinders.