The Infona portal uses cookies, i.e. strings of text saved by a browser on the user's device. The portal can access those files and use them to remember the user's data, such as their chosen settings (screen view, interface language, etc.), or their login data. By using the Infona portal the user accepts automatic saving and using this information for portal operation purposes. More information on the subject can be found in the Privacy Policy and Terms of Service. By closing this window the user confirms that they have read the information on cookie usage, and they accept the privacy policy and the way cookies are used by the portal. You can change the cookie settings in your browser.
Fermentation is always characteristic of multiple nonlinearity and time-varying. In order to know and control the fermentation process, many process models are used to formulate knowledge about process behavior. They are applied, e.g., to predict the process' future behavior and for state estimation when reliable on-line measuring techniques to monitor the key variables of the process are not available...
This paper presents a new image classification method by learning with non-negative matrix factorization (NMF) and SVM. Firstly, NMF is utilized to extract effective features from the high dimensional feature vector. Then the weight coefficients of features are estimated automatically using relevance feedback strategy by linear SVM. NMF and SVM construct a neural network actually. Finally, classification...
In image recognition, feature extraction techniques are widely used to enhance discriminatory performance. In this paper, a new method for image feature extraction, called weighted two-dimensional maximum margin criterion (W2DMMC), is proposed. Different from conventional maximum margin criterion (MMC), W2DMMC is directly based on two-dimensional image matrix rather than one-dimensional vector. And...
This paper proposed a new image Steganalysis scheme based on statistical moments of histogram of multi-level wavelet subbands in frequency domain. Different frequencies of histogram have different sensitivity to various data embedding. Then we decompose the test image using three-level Haar discrete wavelet transform (DWT) into 13 subbands (here the image itself is considered as the LL0 subband).The...
Partial Least Square (PLS) is the most commonly used algorithm for Near Infrared (NIR) modeling. NIR modeling features that it's cheap, easy and fast to measure the NIR spectroscopy, while expensive, difficult and time- consuming to measure the reference value for this spectroscopy. PLS often faces the challenge of that limited samples are available in training set to build a predicative model. To...
Middle -- long forecasting of Electric Power is crucial to the electric investment, which is the guarantee of the healthy development of electric industry. There are so many factors which influence the middle-long electric power load. So in this paper the co-integration technology is used to analyze the influencing factors' inter characters, then a co-integration relating formula is found which is...
We present a type of single-hidden layer feedforward neural networks with the Gaussian activation function. First, we give a new and quantitative proof of the fact that a single layer neural networks with n + 1 hidden neurons can learn n + 1 distinct samples with zero error. Then we give approximate interpolants. They can approximate interpolate, with arbitrary precision, any set of distinct data...
Considering the problem of low classification accuracy of similar category, this paper proposes NFL-SVM, an effective algorithm for solving misclassification in SVM. By analysis and deduction, this paper indicates that SVM can be considered the Nearest Neighbor Classifier of just one feature point in each category. KNFL regards lines of all points as feature lines, so it can be used to overcome defaults...
The least squares support vector machines (LS-SVMs) regression is presented for the purpose of nonlinear dynamic system identification. LS-SVMs are used for system identification of system models with static nonlinear part and dynamic linear part. The actuator saturation is a common nonlinear problem in practical control systems. The identification procedure is illustrated using a simulated example...
For the classical algorithm of BP network model, its convergence rate is slow and it may result in locally optimal solution. But on the condition of same arithmetic complicacy, the Fletcher-Reeves algorithm can improve the convergence rate and come to the least point along the conjugate direction so as to improve the forecasting precision of the BP network model. According to the check results of...
In practices we often expect a fast learning such as real-time or online time series forecasting. However standard algorithms learning the machines from the whole data set are often time consuming. To this end, in this paper we introduce local learning strategy considering only a subset of candidates in the neighborhood of the test point and present a general form of local kernel machines for regression...
Sub-hyper-sphere support vector machines (SVMs) are proposed for solving the classification of the intersections of hyper-spheres when dealing with multi-class classification problem. Since the Gaussian kernel parameter influences the overlap position of the hyper-spheres, the resulting minimum bounding sphere-based classifier must be chosen optimally. This paper presents a new GA-based parameter...
Rock fracture tracing is very important in many rock-engineering applications. This paper presents a new methodology for rock fracture detection, description and classification based on image processing technique and support vector machine (SVM). The developed algorithm uses a number of rock surface images those were taken by sophisticated CCD cameras. The studied algorithm processes all the images...
Considering safety assessment indexes of power supply enterprise are considerable, an hybrid model based on rough sets (RS) and support vector machine(SVM) is proposed: Rough sets, as a anterior preprocessor of SVM, can find out the kernel factors influencing the safety of power supply enterprise by means of attribute reduction algorithm, and then, using them as the input vectors of SVM, the safety...
In this paper, a new short-term traffic flow prediction model and method based on incremental support vector regression (ISVR) is proposed, according to the data collected sequentially by the probe vehicle or loop detectors, which can update the prediction function in real time via incremental learning way. As a result, it is fitter for the real engineering application. The ISVR model was tested by...
Feature selecting for semi-supervised support vector machines (S^3 VM) classifiers is a novel and important research subject in machine learning. For this problem, based on the linear semi-supervised support vector machine (S^3 VM) with 1-norm, we propose two new models using all the available data from labeled and unlabeled data and also utilizing as few of the useful features as possible. Furthermore,...
The resolution of the spectroscopic data can be improved by mathematically removing the degraded effect of the instrument response function. Based on the Shalvi- Weinstein criterion, a statistical neural networks based blind deconvolution algorithm for spectroscopic data is proposed. The true spectra and the spectral slit functions of measure instruments can be estimated simultaneously. Especially,...
Soil erosion is a very complicated process, and influenced by many correlatively factors, so it is hard to evaluate and predict the condition of soil erosion, especially in those regions where there have not sufficiently observation date. To solve the above problem, this paper proposed a new assessment model based on the support vector machines (SVM), In order to improve the accuracy of the model,...
Methods for building near-infrared spectrometry (NIRS) calibration models and for predicting active constituents of rhubarb samples using principal components analysis (PCA) and support vector regression (SVR) were investigated. Principal component analysis was used to reduce the number of spectral variables. Radial basis function neural networks (RBFNNs), ridge regression RBFNNs (RRRBFNNs), and SVR...
Support Vector Machine (SVM), which based on Statistical Learning Theory, is a universal machine learning method. The fault diagnosis of nonlinear and high-controllable High Voltage Direct Current (HVDC) system based on SVM method is proposed, which can take full advantage of effective ability and superiority of SVM in dealing with small samples, and solve many familiar problems in fault diagnosis...
Set the date range to filter the displayed results. You can set a starting date, ending date or both. You can enter the dates manually or choose them from the calendar.