System identification techniques based on support vector machines without bias term

The intention of this article is to utilize support vector machines (SVMs) as process models, which are the basis for most controller designs as well as simulation and monitoring tasks. SVMs are data‐driven models comparable with regularization networks, which merge elements from robust statistics, statistical learning, and kernel theory. The presentation is focused on the ‘no‐bias‐term’ variant, accounts for several peculiarities specific to SVM regression and derives an active‐set algorithm to solve the resulting large‐scale quadratic programming problem. For linear systems, SVMs are combined with multi‐stage methods for estimating output error and ARMAX models. Finally, two real‐world processes serve as test cases to evaluate the SVMs’ properties as nonlinear dynamic models. Copyright © 2013 John Wiley & Sons, Ltd.     

Use of kernel-based Bayesian models to predict late osteolysis after hip replacement

We studied the relationship between osteolysis and polyethylene wear, age at surgery, body mass index and height in 463 subjects (180 osteolysis and 283 controls) after cemented Charnley total hip arthroplasty (THA), in order to develop a kernel-based Bayesian model to quantitate risk of osteolysis. Such tools may be integrated into decision-making algorithms to help personalize clinical decision-making. A predictive model was constructed, and the estimated posterior probability of the implant failure calculated. Annual wear provided the greatest discriminatory information. Age at surgery provided additional predictive information and was added to the model. Body mass index and height did not contain valuable discriminatory information over the range in which observations were densely sampled. The robustness and misclassification rate of the predictive model was evaluated by a five-times cross-validation method. This yielded a 70% correct predictive classification of subjects into osteolysis versus non-osteolysis groups at a mean of 11 years after THA. Finally, the data were divided into male and female subsets to further explore the relationship between wear rate, age at surgery and incidence of osteolysis. The correct classification rate using age and wear rate in the model was approximately 66% for males and 74% for females.     

Effect of sodium carbonate on appearance and textural properties of glutinous rice cake

The effect of sodium carbonate (Na₂CO₃) on the appearance and textural properties of glutinous rice cake was investigated. Glutinous rice kernels were soaked in Na₂CO₃ solution (0–1.2 N) at 25 °C for 24 h. Soaked rice was used for glutinous rice cake (GRC) production. The results showed that the rice soaked in Na₂CO₃ had lower protein and fat content, but higher ash content than those of untreated rice, while the amylose content in treated and untreated rice was not significantly different. Moreover, images from scanning electron microscopy showed that the starch granules from rice treated with Na₂CO₃ had smoother surfaces than untreated ones. From the alkali digestion test, the rice kernels treated with Na₂CO₃ suggest a lower gelatinization temperature than untreated rice. The GRC was uniform, and yellow colour was enhanced. Moreover, GRC exhibited an increase in softness and stickiness with increasing Na₂CO₃ concentration.     

Effect of Pressure on Moisture Transfer During Moisture Adsorption

ABSTRACT

A set of coupled heat, mass, and pressure transfer equations was used to describe the moisture adsorption process in gmn kernels. The finite element method was used to solve the system of equations. The technique was applied to analyze the temperature, moisture, and pressure distribution in a barley kernel during soaking with steep water. The temperature and moisture distributions with (heat, mass, and pressure transfer model) and without (heat and mass uansfer) the effect of pressure were simulated for assumed conditions. The results obtained from the heat, mass, and pressure transfer model show a marked difference from the results obtained from the heat and mass transfer model. This indicated that a pressure gradient exists during the transfer Process, causing additional moisture movement due to filtration effect. Hence. the pressure tern cannot be assumed constant during the moisture adsorption process. The simulated temperature, moisture and pressure profiles and gradients can be used for determining the optimum time required for solking kernels with sleep water to produce barley malt.     

sequential estimation of the hgarginal density function for a strongly mixing process

Suppose that {X_n} is a strongly mixing process with unknow marginal density f(x) and that we estimate f(x) by a kernel estimator [fcirc]_n(x|h_n)and want to achive the MISE no larger than some preassigned postive number w. However,the appropriate sample size n^*depends on a functional of the unknow density function. Therefore some sequential procedure is required and we adopt a fully sequential procedure. In this paper we investigate the asymptotic properties of the procedure and show that the producure is asymptotically efficient in a certain sense as w→0. The results are almost the same in the i.i.d. setting. our result extend a class of models to which the methodology can be applied. For example economic variable,experiments on a single subject in which obervation are not indepent, and so on.     

Parameter-insensitive kernel in extreme learning for non-linear support vector regression

Support vector regression (SVR) is a state-of-the-art method for regression which uses the ε‐sensitive loss and produces sparse models. However, non-linear SVRs are difficult to tune because of the additional kernel parameter. In this paper, a new parameter-insensitive kernel inspired from extreme learning is used for non-linear SVR. Hence, the practitioner has only two meta-parameters to optimise. The proposed approach reduces significantly the computational complexity yet experiments show that it yields performances that are very close from the state-of-the-art. Unlike previous works which rely on Monte-Carlo approximation to estimate the kernel, this work also shows that the proposed kernel has an analytic form which is computationally easier to evaluate.     

A numerical study of vortex interactions with flames developing from ignition kernels in lean methane/air mixtures

In this work, the outcomes of interactions of counter-rotating vortex pairs with developing ignition kernels are studied. The conditions are selected to represent those in a lean-burn natural-gas engine with hot-jet ignition. The evolution of flame surface area during kernel–vortex interaction is quantitatively and qualitatively examined. It is observed that flame development is accelerated and the net flame surface area growth rate, i.e. heat release rate, increased with increasing vortex velocity. In general, increasing the vortex length scale increases the surface growth rate, i.e. increases heat release rates, but for small length scales, i.e. when the ratio of vortex length scale to kernel diameter is small, high flame curvature induced during the interaction leads to flame weakening and slower growth rates. When the vortex velocity is high relative to the flame speed and the length scale is comparable to the kernel diameter, the vortex breaks through the ignition kernel carrying with it hot products of combustion. This accelerates growth of the flame surface area and heat release rates compared to a kernel with no vortex interaction. On decreasing the vortex velocity and increasing the length scale, the wrinkling of the kernel becomes important. This also results in increased surface growth rates and higher heat release rates.     

Response surface optimization and characteristics of rambutan (Nephelium lappaceum L.) kernel fat by hexane extraction

Response surface methodology (RSM) was used to study the effect of moisture content (1.59-18.41 g/100 g), extraction time (2.3-10.7 h) and particle size (0.09-2.11 mm) on the fat yield from rambutan kernels using hexane extraction. The physical and chemical characteristics of rambutan fat were also determined. The optimum conditions obtained from response surface analysis was 4.99 g/100 g moisture, 1.05 mm particle size and 9.2 h extraction time. Under these optimum conditions, the maximum fat yield was 37.35 g/100 g. The extracted fat was a white solid at room temperature. The physical and chemical characteristics of the extracted fat compared well with those of conventional fats. The high level of arachidic acid (∼ 34.3 g/100 g fat) and low iodine value in rambutan kernel fat permits the use of the fat, especially where oxidation may be a concern, without its being subjected to hydrogenation.     

Learning low-rank Mercer kernels with fast-decaying spectrum

Low-rank representations have received a lot of interest in the application of kernel-based methods. However, these methods made an assumption that the spectrum of the Gaussian or polynomial kernels decays rapidly. This is not always true and its violation may result in performance degradation. In this paper, we propose an effective technique for learning low-rank Mercer kernels (LMK) with fast-decaying spectrum. What distinguishes our kernels from other classical kernels (Gaussian and polynomial kernels) is that the proposed always yields low-rank Gram matrices whose spectrum decays rapidly, no matter what distribution the data are. Furthermore, the LMK can control the decay rate. Thus, our kernels can prevent performance degradation while using the low-rank approximations. Our algorithm has favorable in scalability—it is linear in the number of data points and quadratic in the rank of the Gram matrix. Empirical results demonstrate that the proposed method learns fast-decaying spectrum and significantly improves the performance.     

Statistical analysis and adaptive technique for dynamical process monitoring

Multivariate statistical process monitoring (MSPM) methods based on two-dimensional dynamic kernel PCA (2-D-DKPCA) and two-dimensional dynamic kernel Hebbian Algorithm (2-D-DKHA) are proposed. First, a nonlinear batch process monitoring scheme based on 2-D-DKPCA is proposed. Its basic idea is to use KPCA to depict the both within-batch dynamics and batch-to-batch dynamics. However, the proposed 2-D-DKPCA needs to store the whole kernel matrix and calculate all nonlinear components. Kernel matrix will thus become extremely huge when the numbers of successive batches and samples are large. Then, kernel Hebbian Algorithm (KHA) is introduced to 2-D-DKPCA to construct 2-D-DKHA. KHA can extract adaptively nonlinear principal components without storing and manipulating the whole kernel matrix and only calculate the principal components. Thus, proposed 2-D-DKHA has the ability of monitoring complex batch processes. The 2-D-DKPCA and 2-D-DKHA are first proposed in this article. Also, from the proposed 2-D method, it is easily to obtain the 1-D algorithm.