Feature extraction by supervised independent component analysis based on category information

Independent component analysis (ICA) is a technique of transforming observation signals into their unknown independent components; hence, ICA has often been applied to blind signal separation problems. In this application, it is expected that the obtained independent components extract essential information of independent signal sources from input data in an unsupervised fashion. Based on such characteristics, ICA is currently utilized as a feature extraction method for images and sounds for recognition purposes. However, since ICA is an unsupervised learning, the obtained independent components are not always useful in recognition. To overcome this problem, we propose a supervised approach to ICA using category information. The proposed method is implemented in a conventional three‐layered neural network, but its objective function to be minimized is defined not only for the output layer but also for the hidden layer. The objective function consists of the following two terms: one evaluates the kurtosis of hidden unit outputs and the other evaluates the error between output signals and their teacher signals. The experiments are performed using several standard datasets to evaluate performance of the proposed algorithm. It is confirmed that a higher recognition accuracy is attained by the proposed method as compared with a conventional ICA algorithm. © 2007 Wiley Periodicals, Inc. Electr Eng Jpn, 161(2): 25–32, 2007; Published online in Wiley InterScience ( www.interscience.wiley.com ). DOI 10.1002/eej.20522     

Monte Carlo simulation of hydrogen absorption in palladium and palladium–silver alloys

A modified Monte Carlo (MC) simulation was performed to investigate the hydrogen absorption behavior in Pd and Pd–Ag alloys of the composition Pd_xAg_1−x (x=0.7–0.8) under H₂ pressure (0.1 MPa) at different temperatures. The present method employed can consider the dissociative adsorption of hydrogen molecule and the subsequent absorption of hydrogen atom by formalizing the relationship between the pressure of hydrogen molecule and hydrogen atom. The potential parameters were determined to reproduce the solution enthalpy of hydrogen in pure metals. The results are in good agreement with experimental findings as well as previous theoretical studies. We confirmed that our method is useful to simulate the absorption of hydrogen in metals and alloys.     

Hebbian learning rule restraining catastrophic forgetting in pulse neural network

In this paper, a Hebbian learning rule restraining “catastrophic forgetting” is proposed on a pulsed neural network (PNN) with leaky integrate‐and‐fire neurons. The strong point of this learning rule is that a learning of new pattern does not destroy past ones, and that an efficient use of synapses is enabled. First, in order to consider the function of the learning rule, a fundamental experiment is carried out. Next, to compare the performance between the proposed learning rule and conventional ones on the application, simulation experiments are examined using autonomous behavior robots which are forced to learn concurrently two different environments. The results of the experiments show that the proposed learning rule clearly restrains “catastrophic forgetting” and enables working of more efficient than conventional PNN learning. © 2005 Wiley Periodicals, Inc. Electr Eng Jpn, 151(3): 50–60, 2005; Published online in Wiley InterScience ( www.interscience.wiley.com ). DOI 10.1002/eej.10343     

Characterization of interfacial reaction and chemical bonding features of LaOx/HfO2 stack structure formed on thermally-grown SiO2/Si(1 0 0)

A stack structure consisting of ～1.5 nm-thick LaO_x and ～4.0 nm-thick HfO₂ was formed on thermally grown SiO₂ on Si(1 0 0) by MOCVD using dipivaloymethanato precursors, and the influence of N₂ annealing on interfacial reaction for this stack structure was examined by using X-ray photoelectron spectroscopy and Fourier transform infrared attenuated total reflection. We found that compositional mixing between LaO_x and HfO₂ becomes significant from 600 °C upwards and that interfacial reaction between HfLa_yO_z and SiO₂ proceeds consistently at 1000 °C in N₂ ambience.     

Saturation measurements on electron spin resonance signals in amorphous silicon films

The dependence of the electron spin resonance signal intensity on the microwave magnetic field was investigated for amorphous silicon prepared by r.f. sputtering. The spin-lattice and the spin relaxation times T₁ and T₂ were determined by a new saturation method. As a result it is suggested that the electron spin resonance linewidth is determined by g-factor anisotropy, that T₂ is 10–100 times longer than the T¹₂ determined from the linewidth, that T₁ for samples annealed at below about 250°C is proportional to T^-1.9 and that most of the spins for samples annealed at above 400°C are probably due to dangling bonds in amorphous regions surrounding crystallites newly produced in the process of crystallization.     

Automatic welding of pipeline in Japan: Part 2—The development of a method for the branch pipe joint in ship pipeline

The pipelines used in a ship amount to tens of thousands of metres per ship, and so automatic welding of the pipelines has been widely applied from the viewpoint of saving labour. In most cases the automatic welding process has been used for the welding of butt-joints which can be applied to the pipes in rotary motion, and it has also been used for fillet-welding of the connection between flange and piping. In the butt-welding of fixed pipings, automatic welding processes are now being partly used for position welding but up to the present time automatic welding has seldom been applied to the various types of branch pipe joints in which the welded lines form three-dimensional curves.In recent years, great progress has been made in the technique of constructing steel pipe structures and in the development of practical gas cutting processes to prepare the curves needed for the various welded joints used in piping. This paper deals with various conditions and performance requirements needed for automatic welding, such as the selection of shapes of prepared edges of pipings to be welded, the control of welding positions, the control of welding torches in operation, and the solution and control of other production variables.     

The Application of Marginal Analysis to the Economic Lot Scheduling Problem

Application of marginal analysis to the Economic Lot Scheduling Problem reduces the solution search to a simple systematic calculation and feasibility check. Compared with mathematical programming methods and heuristic approaches advocated previously, marginal analysis results in lower optimum cost solution with shorter optimum basic period lengths. For comparative purposes, data are included which allow evaluation of marginal analysis results with dynamic programming results of previous researchers.     

Molecular logic and computing

Molecular substrates can be viewed as computational devices that process physical or chemical 'inputs' to generate 'outputs' based on a set of logical operators. By recognizing this conceptual crossover between chemistry and computation, it can be argued that the success of life itself is founded on a much longer-term revolution in information handling when compared with the modern semiconductor computing industry. Many of the simpler logic operations can be identified within chemical reactions and phenomena, as well as being produced in specifically designed systems. Some degree of integration can also be arranged, leading, in some instances, to arithmetic processing. These molecular logic systems can also lend themselves to convenient reconfiguring. Their clearest application area is in the life sciences, where their small size is a distinct advantage over conventional semiconductor counterparts. Molecular logic designs aid chemical (especially intracellular) sensing, small object recognition and intelligent diagnostics.