Extensions of LDA by PCA mixture model and class-wise features

Linear discriminant analysis (LDA) is a data discrimination technique that seeks transformation to maximize the ratio of the between-class scatter and the within-class scatter. While it has been successfully applied to several applications, it has two limitations, both concerning the underfitting problem. First, it fails to discriminate data with complex distributions since all data in each class are assumed to be distributed in the Gaussian manner. Second, it can lose class-wise information, since it produces only one transformation over the entire range of classes. We propose three extensions of LDA to overcome the above problems. The first extension overcomes the first problem by modelling the within-class scatter using a PCA mixture model that can represent more complex distribution. The second extension overcomes the second problem by taking different transformation for each class in order to provide class-wise features. The third extension combines these two modifications by representing each class in terms of the PCA mixture model and taking different transformation for each mixture component. It is shown that all our proposed extensions of LDA outperform LDA concerning classification errors for synthetic data classification, hand-written digit recognition, and alphabet recognition.     

High Gd Concentration GaGdN Grown at Low Temperatures

GaGdN layers were grown at temperatures below 300°C by radio-frequency plasma-assisted molecular beam epitaxy on sapphire substrates. GaGdN samples with high Gd concentration as high as 12.5% were obtained by lowering the growth temperature. X-ray diffraction results showed no obvious secondary phase, which means that the phase separation can be suppressed by the growth at low temperatures. All samples, including those grown at room temperature, showed ferromagnetic characteristics. Photoluminescence emission was observed, though spectra exhibit broad and sharp luminescence bands related to many kinds of defects. It is suggested that electrons coming from defects, especially, nitrogen vacancy, stabilize ferromagnetism, and that the carrier-induced ferromagnetism occurs in the low-temperature-growth GaGdN.     

Trenched‐Sinker LDMOSFET (TS‐LDMOS) Structure for 2 GHz Power Amplifiers

This paper proposes a new LDMOSFET structure with a trenched sinker for high‐power RF amplifiers. Using a low‐temperature, deep‐trench technology, we succeeded in drastically shrinking the sinker area to one‐third the size of the conventional diffusion‐type structure. The RF performance of the proposed device with a channel width of 5 mm showed a small signal gain of 16.5 dB and a maximum peak power of 32 dBm with a power‐added efficiency of 25% at 2 GHz. Furthermore, the trench sinker, which was applied to the guard ring to suppress coupling between inductors, showed an excellent blocking performance below ?40 dB at a frequency of up to 20 GHz. These results confirm that the proposed trenched sinker should be an effective technology both as a compact sinker for RF power devices and as a guard ring against coupling.     

Threshold‐Based Camera Motion Characterization of MPEG Video

We propose an efficient scheme for camera motion characterization in MPEG‐compressed video. The proposed scheme detects six types of basic camera motions through threshold‐based qualitative interpretation, in which fixed thresholds are applied to motion model parameters estimated from MPEG motion vectors (MVs). The efficiency and robustness of the scheme are validated by the experiment with real compressed video sequences.     

Computation of the average power pattern of a reflector antennawith random surface errors and misalignment errors

To specify manufacturing tolerances of a reflector antenna, various errors such as random surface errors and misalignment errors must be considered at one time because superposition of the effects of those errors may not hold. Based on the Rahmat-Samii's formulation (1983), a method for computing efficiently the average power pattern of a reflector antenna with those errors is presented. Simulation results show that superposition of the effects of errors does not generally hold and demonstrate how those errors degrade the peak-gain and sidelobe levels     

An efficient assumed strain element model with six DOF per node for geometrically non-linear shells

The present paper describes an assumed strain finite element model with six degrees of freedom per node designed for geometrically non-linear shell analysis. An important feature of the present paper is the discussion on the spurious kinematic modes and the assumed strain field in the geometrically non-linear setting. The kinematics of deformation is described by using vector components in contrast to the conventional formulation which requires the use of trigonometric functions of rotational angles. Accordingly, converged solutions can be obtained for load or displacement increments that are much larger than possible with the conventional formulation with rotational angles. In addition, a detailed study of the spurious kinematic modes and the choice of assumed strain field reveals that the same assumed strain field can be used for both geometrically linear and non-linear cases to alleviate element locking while maintaining kinematic stability. It is strongly recommended that the element models, described in the present paper, be used instead of the conventional shell element models that employ rotational angles.     

The design of robots and intelligent manipulators using modern composite materials

The operating speed and endpoint positional accuracy of existing industrial manipulators are limited by the inertial and stiffness characteristics of the articulating members of the robot's mechanical linkage. This limitation may be overcome by developing members having high structural stiffness and strength with low mass, and this has been recognized for some time. These characteristics can be obtained by fabricating the moving members of manipulators in fiber reinforced composite materials. In order to establish a basis for the dynamic analysis of robots fabricated in viscoelastic composites, a variational theorem is developed herein. A preliminary comparative study is then undertaken for manipulators manufactured in a graphite-epoxy composite material and also steel in order to demonstrate some of the advantages to be accrued from this proposed new design philosophy.     

Application of differential quadrature to static analysis of structural components

The numerical technique of differential quadrature for the solution of linear and non-linear partial differential equations, first introduced by Bellman and his associates, is applied to the equations governing the deflection and buckling behaviour of one- and two-dimensional structural components. Separate transformations are used for higher-order derivatives, as suggested by Mingle, thus extending the method to treat fourth-order equations and to include multiple, boundary conditions in the respective co-ordinate directions. Results are obtained for various boundary and loading conditions and are compared with existing exact and numerical solutions by other methods. The application of differential quadrature to this class of problems is seen to lead to accurate results with relatively small computational effort.     

Effect of bulk flow pulsations on film cooling with compound angle holes

Experiments are conducted to investigate the effects of bulk flow pulsations on film cooling with compound angle holes. A row of five film cooling holes is considered with orientation angles of 0°, 30°, 60°, and 90° at a fixed inclination angle of 35°. Static pressure pulsations are produced by an array of six rotating shutter blades, which extend across the span of the exit of the wind tunnel test section. The pulsation frequency is fixed at 36 Hz, but changes in the time-averaged blowing ratios of 0.5, 1.0 and 2.0 produce three different coolant Strouhal numbers, 3.6, 1.8 and 0.9, respectively. Detailed film cooled boundary layer temperature distributions are measured by a cold wire and the adiabatic film cooling effectiveness by thermochromic liquid crystal (TLC). The boundary layer temperature surveys show that pulsations induce large disruptions to the boundary layer temperature distribution and the film coverage. As the orientation angle increases, the injectant concentration spreads further into the spanwise direction because of pulsations than the steady case. With pulsations the adiabatic film cooling effectiveness value decreases regardless of the orientation angle. The amount of reduction, however, depends on the orientation angle in such a way that the larger the orientation angle is, the smaller the reduction is.     

A lagrangian model for spray dryer desulfurization system

The purpose of this work is to develop a reliable numerical model for the spray dryer desulfurization system. The shape of the spray dryer requires that a body fitted orthogonal coordinate system be used for the numerical model. The governing equations are developed in the general orthogonal coordinates and discretized to yield a system of algbraic equations. A modified two-equation turbulence model is also included in the numerical program. The stability of QUICK scheme is investigated and the Absolutely Stable QUICK scheme is developed and included in the numerical model. The trajectory approach is used to simulate the flow of the dispersed phase. Two-way coupling phenomena is modeled by this sheme. The absorption of sulfur dioxide into lime slurry droplets is simulated by a model based on gas-phase mass transfer. The program is applied to a typical spray dryer desulfurization system. The reuslts show the capability of the program to predict the sensitivity of system performance to change in operational parameters.