Papers by Keyword: Support Vector Regression (SVR)

Paper TitlePage

Authors: Bi Qiang Yao
Abstract: The availability of efficient and accurate metamodel for optimization computation is crucial to the success of applications of robust optimization of computationally intensive simulation models. To address this need, a framework has been presented for robust optimization on problems that involve high dimensional. The framework was combined with support vector regression (SVR) approximation model and a genetic algorithm (GA). The performances of SVR were compared with those of polynomial regression (PR), Kriging and back-propagation neural networks (BPNN). The results showed that the prediction accuracy of SVR model was higher than those of others metamodels. The applicability of the algorithm developed by combining SVR and GA was demonstrated by using a two-bar structure system study, and was found to be accurate and efficient for robust optimization. The optimization framework was effectively utilized to achieve a potential performance improvement.
Authors: Tong Di He, Jian Wei Li
Abstract: In order to improve water quality retrievals of multi-spectral image accurately, this paper puts forward a method for water quality remote retrieva based on support vector regression with parameters optimized by genetic algorithm. The method uses SPOT-5A data and the water quality field data, chose four representative water quality parameters, support vector regression are trained and tested, the parameters of support vector regression are optimized by genetic algorithms. The result of experiment shows that the method has more accuracy than the routine method. It provides a new approach for remote sensing monitoring of environment.
Authors: De Huai Zeng, Yuan Liu, Lian Bo Jiang, Li Li, Gang Xu
Abstract: In this paper, support vector regression with ant colony optimization is presented for the prediction of tool-chip interface temperature depends on cutting parameters in machining. Ant colony (ACO) optimization was developed to optimize three parameters of SVR, including penalty parameter C, insensitive loss function ε and kernel function σ. SVR constructs hyperplane in high dimension space and fits the data in non-linear form. Normalized mean square error (NMSE) of fitting result is used as target of ant colony optimization. ACO finds the best parameters which correspond to the NMSE. The results showed that the proposed approach, by comparing with back-propagation neural network model, was an efficient way to model tool–chip interface temperature with good predictive accuracy.
Authors: Shao Chao Sun, Dao Huang
Abstract: In this paper, we propose a new type of ε-insensitive loss function, called as ε-insensitive Fair estimator. With this loss function we can obtain better robustness and sparseness. To enhance the learning speed ,we apply the smoothing techniques that have been used for solving the support vector machine for classification, to replace the ε-insensitive Fair estimator by an accurate smooth approximation. This will allow us to solve ε-SFSVR as an unconstrained minimization problem directly. Based on the simulation results, the proposed approach has fast learning speed and better generalization performance whether outliers exist or not.
Authors: Qing Wu
Abstract: This paper presents a new smooth approach to solve support vector regression (SVR). Based on Karush-Kuhn-Tucker complementary condition in optimization theory, a smooth unconstrained optimization model for SVR is built. Since the objective function of the unconstrained SVR model is non-smooth, we apply the smooth techniques and replace the ε-insensitive loss function by CHKS function. Newton-Armijo algorithm is used to solve the smooth CHKS-SSVR model. Primary numerical results illustrate that our proposed approach improves the regression performance and the learning efficiency.
Authors: Xiao Ming Xue, Jian Zhong Zhou, Yong Chuan Zhang, Xiao Jian, Xue Min Wang
Abstract: The end effects is a serious problem in the applications of the empirical mode decomposition (EMD) method. To deal with this problem, an extrema extension method based on the support vector regression (SVR) is proposed in this paper. In each iterating process of the EMD method, the SVR method is employed to predict one maximum and a minimum point respectively at the both ends of the original data series to form the relatively true upper and lower envelope, thus the end effects can be restrained effectively. The prediction of an extrema point includes two parts, the forecast of the extreme value and location. In contrast with other traditional extrema extension methods, such as the extrema mirror extension and linear fitting extension method, the decomposed results from the simulation and actual signals demonstrated that this proposed method has a better performance in eliminating the end effects related to the empirical mode decomposition.
Authors: De Huai Zeng, Yuan Liu, Li Li, De Gui Yu, Gang Xu
Abstract: With the development of high power LED technology, junction temperature as a key factor constrains the performance and the service life of LED, and the main parameter of junction temperature is thermal resistance. Therefore, how to measure the thermal resistance of high power LED quickly and accurately plays an important part in improving the performance and the service life of LED. In this paper the accurate and fast measurement equipment was applied to study the thermal characteristics of high power LED. The forward-voltage based method was conducted to measure the junction temperature of high power. Then, support vector regression (SVR) combined with genetic algorithm (GA) for its parameter optimization, was proposed to establish a model to predict the thermal resistance of high power LED. The prediction performance of GA-SVR was compared with those of BPNN model. The result demonstrated that the estimated errors GA-SVR models, such as Mean Absolute Relative Error (MARE) and Root Mean Squared Errors (RMSE), all are smaller than those achieved by the BPNN applying identical samples.
Authors: Jie Fang Liu
Abstract: Support vector machine (SVM) is based on the principle of structural risk minimization, which makes SVM has better generalization ability than other traditional learning machines that are based on the learning principle of empirical risk minimization.Research on the application of Support vector regression (SVR) model in spectrophotometry was done to determine the content of benzoic acid and salicylic acid simultaneously. The predicted result was found highly correlated with the time when the data was collected to build the model. The closer of the dates between collecting data for modeling and for predicting, the better the predicted results. SVR model with significantly improved robustness was resulted by using all the collected data over time, which, when applied to the determination of benzoic acid and salicylic acid simultaneously, led to satisfactory result, with recoveries being 97%-102%.
Authors: Xin Ruo Hua
Abstract: The common determination method of pile point and pile bearing capacity always depends on the practical experiences. The accuracy of pile survey data should be verified in the field. To predict the pile point and pile bearing capacity, a prediction model based on support vector regression algorithms has been developed. The friction resistance around piles and the static-load settlement are the inputs. The pile point bearing capacity characteristic value and the characteristic value of pile bearing capacity are the outputs. The prediction model is trained by the data in the adjacent block and validated by the field data in the project. The calculations of 40mm settlement and the prediction of other piles in the project can then be acquired. Compared to the survey data, the predicted values has 16% surplus. This will be beneficial to reduce construction work significantly and to save about 5% foundation cost.
Authors: Xue Mei Li, Li Xing Ding, Jin Hu Lǔ, lan Lan Li
Abstract: Accurate forecasting of building cooling load has been one of the most important issues in the electricity industry. Recently, along with energy-saving optimal control, accurate forecast of electricity load has received increasing attention. Because of the general nonlinear mapping capabilities of forecasting, artificial neural networks have played a crucial role in forecasting electricity load. Support vector machines (SVMs) have been successfully employed to solve nonlinear regression and time series problems. In order to improve time efficiency of prediction, a new hourly cooling load prediction model and method based on Support Vector Machine in this paper. Moreover, simulated annealing (SA) algorithms were employed to choose the parameters of a SVM model. Subsequently, examples of cooling load data from Guangzhou were used to illustrate the proposed SVM-SA model. A comparison of the performance between SVM optimized by Particle Swarm Optimization (SVM-PSO) and SVM-SA is carried out. Experiments results demonstrate that SVM-SA can achieve better accuracy and generalization than the SVM-PSO. Consequently, the SVM-SA model provides a promising alternative for forecasting building load.
Showing 1 to 10 of 50 Paper Titles