Changjiang Zhang

Prediction of Chaotic Time Series Using LS-SVM with Automatic Parameter Selection

Least squares support vector machine (LS-SVM) combined with genetic algorithm (GA) is used to predict chaotic time series. The LS-SVM can overcome some shortcoming in the multilayer perceptron and the GA is used to tune the LS-SVM parameters automatically. A benchmark problem, Hénon map time series, has been used as an example for demonstration. It is showed this approach can escape from the blindness of man-made choice of the LS-SVM parameters. It enhances the efficiency and the capability of prediction. Further, the GA is compared with cross-validation method for tuning LS-SVM parameters. The results reveal that the GA can obtain lower prediction errors than the k-folds cross validation method.

Application of Adaptive Least Square Support Vector Machines in Nonlinear System Identification

Training problem of least squares support vector machine (LS-SVM) is solved by finding a solution to a set of linear equations. This makes online adaptive implementation of the algorithm feasible. In this paper, an adaptive algorithm for the purpose of nonlinear system identification is proposed. Using this training algorithm, a variant of support vector machine has been developed called adaptive LS-SVM. The adaptive LS-SVM is especially useful on online system identification. Several pertinent numerical simulations have shown the validity of the proposed method

A Reducing Multi-Noise Contrast Enhancement Algorithm for Infrared Image

A kind of infrared image contrast enhancement algorithm based on discrete stationary wavelet transform (DSWT) and nonlinear gain operator is proposed. Having implemented DSWT to an infrared image, de-noising is done by the method proposed in the high frequency sub-bands which are in the better resolution levels and enhancement is implemented by combining de-noising method with incomplete Beta transform (IBT) in the high frequency sub-bands which are the worse resolution levels. According to experimental results, the new algorithm can reduce effectively the correlative noise (1/f noise), additive Gauss white noise (AGWN) and multiplied noise (MN) in the infrared image while it also enhances the contrast of infrared image well. In visual quality, the algorithm is better than the traditional unshaped mask method (USM), histogram equalization method (HIS)

Signals recognition of electronic nose based on support vector machines

A new intelligent method for signals recognition of electronic nose, based on support vector machine (SVM) classification, is presented. The SVM operates on the principle of structure risk minimization; hence a better generalization ability is guaranteed. This paper discusses the basic principle of the SVM at first, and then uses it as a classifier to recognize the gas category. The method can classify complicated patterns and achieve higher recognition rate at reasonably small size of training sample set and can overcome disadvantages of the artificial neural networks. The experiments of the recognition of three different gases, ethanol, gasoline and acetone, have been presented and discussed. The results indicate that the SVM classifier exhibits good generalization performance and enables the average recognition rate to reach 88.33% for the testing samples. This means the method proposed is effective for signals recognition of electronic nose.

Time series prediction using LS-SVM with particle swarm optimization

Time series analysis is an important and complex problem in machine learning. In this paper, least squares support vector machine (LS-SVM) combined with particle swarm optimization (PSO) is used to time series prediction. The LS-SVM can overcome some shortcoming in the multilayer perceptron (MLP) and the PSO is used to tune the LS-SVM parameters automatically. A benchmark problem, Henon map time series, has been used as an example for demonstration. It is showed this approach can escape from the blindness of man-made choice of the LS-SVM parameters. It enhances the efficiency and the capability of prediction.

An efficient non-linear algorithm for contrast enhancement of infrared image

An infrared image contrast enhancement algorithm based on a discrete stationary wavelet transform (DSWT) and non-linear gain operator is proposed. Having implemented DSWT on an infrared image, de-noising is done using the method proposed in high frequency sub-bands which have better resolution levels and enhancement is implemented by combining the de-noising method with the non-linear gain method in high frequency sub-bands which have the worse resolution levels. According to experimental results, the new algorithm can reduce effectively the correlative noise (1/f noise), additive Gauss white noise (AGWN) and multiplicative noise (MN) in the infrared image while it also enhances contrast of the infrared image. In terms of visual quality, the algorithm is better than the traditional unshaped mask method (USM), histogram equalization method (HIS) and two methods by Gong et al. (2000) and Wu et al. (2003).

A New Support Vector Machine and Its Learning Algorithm

Support vector machine is a learning technique based on the structural risk minimization principle, this paper proposes a new kind of support vector machine (SVM), which modifies the classical SVM formulation to get even simpler dual optimization problem, then gives a quadratic optimization theorem, and according to it derives a multiplicative updates algorithm for solving the dual optimization problem. The updates algorithms converge monotonically to the solution of the optimal problem, and have a simple closed form. Experimental results of simulation indicate the feasibility of the varied regression support vector machine and its training algorithm

PQ Disturbances Identification Based on SVMs Classifier

The deregulation polices in electric power systems result in the absolute necessity to quantify power quality (PQ). An effective classification strategy for PQ disturbances was needed. A new method based on N-I support vector machines (SVMs) was presented for PQ disturbances identification. Through phase-shift and some simple algebra operations, the PQ disturbances were detected first. Then a data dealing process was carried out to extract features from the detecting outputs. Then N kinds of PQ disturbances were classified with an N-I SVMs classifier. The testing results show that the proposed method could classify the PQ disturbances successfully. Moreover, the classifier has an excellent performance on training speed and reliability

Modeling nonlinear dynamical systems using support vector machine

This paper proposes a general framework for modeling nonlinear dynamical systems based on support vector machine (SVM), firstly provides a short introduction to regression SVM, then uses standard support vector machine to model nonlinear dynamical system, and gives a theoretic analysis about its robustness under noise. The simulation results indicate that the SVM method can reduce the effect of samples number and noise for modeling, and its performance is better than that of neural network modeling method.

An efficient de-noising algorithm for infrared image

Employing discrete stationary wavelet transform (DSWT) and generalized cross validation (GCV), an efficient denoising algorithm for infrared image is proposed. Asymptotical optimal threshold can be obtained, without knowing the variance of noise, only employing the known input image data. Having implemented DSWT to an infrared image, additive Gauss white noise (AGWN), 1/f noise and multiplicative noise (MN) can be suppressed efficiently in the high frequency sub-bands of each decomposition level respectively. Experimental results show that the new algorithm can reduce efficiently the AGWN and 1/f noise in the infrared image while keeps the detail information of targets well. In performance index and visual quality, the new algorithm is more excellent than the de-noising algorithm based on discrete orthogonal wavelet transform (DOWT) and the conditional median value filter (MVF).