Improving visualisation and prediction performance of supervised self-organising map by modified contradiction resolution

We have previously proposed a new type of information-theoretic method where a neuron is evaluated by itself (self-evaluation) and by its surrounding neurons (outer-evaluation). If contradiction between different types of evaluation exists, it is reduced as much as possible. In the present paper, we try to separate self- and outer-evaluation more explicitly and introduce the importance of neurons. First, we separate self- and outer-evaluation to enhance the characteristics shared by the two types of evaluation. Second, we introduce the importance of neurons in evaluation. By using a limited number of important neurons in evaluation, we expect the main characteristics in input patterns to emerge. We applied this contradiction resolution to two types of data, namely, the Senate data and the Euro-yen exchange rates. In both data sets, experimental results confirmed that improved prediction performance was obtained. Prediction performance was better than that obtained by the conventional self-organising map (SOM) and radial basis function networks. In addition, final representations obtained by contradiction resolution were easier to interpret than those given by the conventional SOM. Experimental results confirmed that improved interpretation and visualisation were accompanied by improved prediction performance.     

New approach for synthesis of activated carbon from bamboo

Unconventional pretreatment, that is, delignification and the addition of guanidine phosphate, was performed for the synthesis of activated carbon having a high specific surface area from bamboo by physical activation. The values of the specific surface area, total pore volume, and average pore size depended on the amount of added guanidine phosphate and the CO₂ activation time. The amount of the added guanidine phosphate under the optimum conditions for the highest specific surface area was much lower than that of the phosphorous acid chemical activator under conventional conditions. The N₂ adsorption isotherms of all the samples were type I, which means that micropores were dominant. The pore sizes of the samples in this study were similar to that of the physically-activated carbon. Therefore, the activation process was presumed to be essentially not chemical, but physical. The relation between the yield and the specific surface area improved with the addition of guanidine phosphate. The reason for the improvement may be the change in the reactivity of the carbon material generated during the heating process. The maximum specific surface area was ca. 2000 m² g^?1, which is a high value for a physically-activated carbon.     

Flexible feature discovery and structural information control

In this paper, we propose a new information theoretic method called structural information control for flexible feature discovery. The new method has three distinctive characteristics, which traditional competitive learning fails to offer. First, the new method can directly control competitive unit activation patterns, whereas traditional competitive learning does not have any means to control them. Thus, with the new method, it is possible to extract salient features not discovered by traditional methods. Second, competitive units compete witheach other by maximizing their information content about input patterns. Consequently, this information maximization makes it possible to control flexibly competition processes. Third, in structural information control, it is possible to define many different kinds of information content, and we can choose a specific type of information according to a given objective. When applied to competitive learning, structural information can be used to control the number of dead or spare units, and to extract macro as well as micro features of input patterns in explicit ways. We first applied this method to simple pattern classification to demonstrate that information can be controlled and that different neuron firing patterns can be generated. Second, a dipole problem was used to show that structural information could provide representations similar to those by the conventional competitive learning methods. Finally, we applied the method to a language acquisition problem in which networks must flexibly discover some linguistic rules by changing structural information. Especially, we attempted to examine the effect of the information parameter to control the number of dead neurons, and thus to examine how macro and micro features in input patterns can explicitly be discovered by structural information.     

Input information maximization for improving self-organizing maps

In this paper, we propose a new type of information-theoretic method called “input information maximization” to improve the performance of self-organizing maps. We consider outputs from input neurons by focusing on winning neurons. The outputs are based on the difference between input neurons and the corresponding winning neurons. Then, we compute the uncertainty of input neurons by normalizing the outputs. Input information is defined as a decrease in the uncertainty of input neurons from a maximum and observed value. When input information increases, fewer input neurons tend to be activated. In the maximum state, only one neuron is on, and all others are off. We applied the method to two data sets, namely, the Senate and voting attitude data sets. In both, experimental results confirmed that when input information increased, quantization and topographic errors decreased. In addition, clearer class structure could be extracted by increasing input information. In comparison to our previous methods to detect the importance of input neurons, the present method turned out to be good at producing faithful representations with much more simplified computational procedures.     

Experimental evaluation of the Peierls stresses in a variety of crystals and their relation to the crystal structure

Peierls stresses (τ_P) of dislocations of 66 slip systems in 52 crystals were estimated from experimental data either by direct extrapolation of the critical resolved shear stress (τ_c) vs. temperature curve to absolute zero temperature, or from the T₀ value at which the temperature dependence of τ_c vanishes, based on the kink-pair formation enthalpy, which is a function of τ_P, described by the line tension model of the dislocation. The normalized τ_P/G (G is the shear modulus) values are distributed over four orders of magnitude, but τ_P/G values for a group of crystals with the same crystal structure are within an order of magnitude, indicating a homologous nature of τ_P in crystals. In order to compare the results with the Peierls–Nabarro (P–N) formula generalized to any dislocation character, log(τ_P/G) values were correlated with crystal parameters. In this generalized P–N plot, most of the plots deviate downwards from the Huntington relation (a revised, original P–N relation) and the results of the discretized P–N models of Ohsawa et al. (Ohsawa K, Koizumi H, Kirchner HOK, Suzuki T. Philos Mag A 1994;69:171), with the deviation becoming larger at large h/δ value, where h is the lattice spacing of the glide plane, and δ is the period of the lattice in the direction of the dislocation glide. In the plot, there is a tendency that the stronger the covalent character, the higher the τ_P/G value, reflecting the general tendency of the normalized theoretical shear strength of crystals.     

Analysis of a gyroscopic force measuring system in three-dimensional space

This paper concerns the development of an entirely new sensor for measuring forces using a gyroscope (called gyroscopic force measuring system, or simply GFMS) for measuring a force vectorially. In a previous paper [S. Kurosu, M. Adachi, K. Kamimura, Dynamical characteristics of gyroscopic weight measuring device, ASME J. Dynamic System Measurement, and Control 119 (1997) 346–350], the dynamics of the GFMS for measuring a vertical force were investigated and the principal characteristics were examined theoretically and experimentally. The results of this work are directly applicable to measurement of a vectorial force in three-dimensional space. The principle and the dynamical characteristics of the GFMS for measuring a force vector are analyzed theoretically. To measure a force vectorially, two auxiliary turntables (driven by servomechanisms) are installed around the gyroscope, in which turntable outputs are required to follow some angles of incidence of a force vector. Some unfavorable errors caused by various factors and disturbances are analyzed. Two types of compensation methods are proposed as a device both for accurate force measurement and disturbance suppression. The feasibility of the proposed GFMS is confirmed by numerical simulations.     

Wien2wannier: From linearized augmented plane waves to maximally localized Wannier functions

We present an implementation of an interface between the full-potential linearized augmented plane wave package Wien2k and the wannier90 code for the construction of maximally localized Wannier functions. The FORTRAN code and a documentation is made available and results are discussed for SrVO₃, Sr₂IrO₄ (including spin-orbit coupling), LaFeAsO, and FeSb₂.     

A 144-Mb, eight-level NAND flash memory with optimized pulsewidthprogramming

We report a fast-programming, compact sense and latch (SL) circuit to realize an eight-level NAND flash memory. Fast programming is achieved by supplying optimized voltage and pulsewidth to the bit lines, according to the programming data. As a result, all data programming is completed almost simultaneously, and 0.67-MB/s program throughput, which is 1.7 times faster than conventional program throughput, is achieved. The compact layout of the SL circuit is made possible by four 3-bit latches sharing one unit of the read/verify control circuit. Using these techniques, we fabricated a 144-Mb, eight-level NAND flash memory using a 0.35-μm CMOS process, resulting in a 104.2-mm² die size and a 1.05-μm² effective cell size