SVM在基因表达数据分类中的研究和应用

介绍了一种使用基因芯片实验产生的基因表达数据对功能基因进行分类的方法,该方法是以支持向量机(SVM)理论为基础的。文中描述了径向基函数SVM,与其它SVM相比,径向基函数SVM在基因分类中有更好的性能。SVM的理论基础是统计学习理论,它不仅结构简单,而且技术性能高,泛化能力强,在基因表达式分类中表现出有很多优点,成为热点研究方向。

Support vector machines approach to mean particle size of rock fragmentation due to bench blasting prediction

Aiming at the problems of the traditional method of assessing distribution of particle size in bench blasting, a support vector machines （SVMs） regression methodology was used to predict the mean particle size （X50） resulting from rock blast fragmentation in various mines based on the statistical learning theory. The data base consisted of blast design parameters, explosive parameters, modulus of elasticity and in-situ block size. The seven input independent variables used for the SVMs model for the prediction of X50 of rock blast fragmentation were the ratio of bench height to drilled burden （H/B）, ratio of spacing to burden （S/B）, ratio of burden to hole diameter （B/D）, ratio of stemming to burden （T/B）, powder factor （Pf）, modulus of elasticity （E） and in-situ block size （XB）. After using the 90 sets of the measured data in various mines and rock formations in the world for training and testing, the model was applied to 12 another blast data for validation of the trained support vector regression （SVR） model. The prediction results of SVR were compared with those of artificial neural network （ANN）, multivariate regression analysis （MVRA） models, conventional Kuznetsov method and the measured X50 values. The proposed method shows promising results and the prediction accuracy of SVMs model is acceptable.

Predicting pillar stability for underground mine using Fisher discriminant analysis and SVM methods

The purpose of this study is to apply some statistical and soft computing methods such as Fisher discriminant analysis （FDA） and support vector machines （SVMs） methodology to the determination of pillar stability for underground mines selected from various coal and stone mines by using some index and mechanical properties, including the width, the height, the ratio of the pillar width to its height, the uniaxial compressive strength of the rock and pillar stress. The study includes four main stages： sampling, testing, modeling and assessment of the model performances. During the modeling stage, two pillar stability prediction models were investigated with FDA and SVMs methodology based on the statistical learning theory. After using 40 sets of measured data in various mines in the world for training and testing, the model was applied to other 6 data for validating the trained proposed models. The prediction results of SVMs were compared with those of FDA as well as the measured field values. The general performance of models developed in this study is close; however, the SVMs exhibit the best performance considering the performance index with the correct classification rate Prs by re-substitution method and Pcv by cross validation method. The results show that the SVMs approach has the potential to be a reliable and practical tool for determination of pillar stability for underground mines.

GRADE-LIFE PROGNOSTIC MODEL OF AIRCRAFT ENGINE BEARING

Research on practical and verifiable prediction methods for the service life of bearings plays a critical role in improving the reliability and safety of aircraft engines. The concept of grade-life （GL） is introduced to de- scribe the service life of bearings. A GL prognostic model for aircraft engine bearings is proposed based on sup- port vector machine （SVM） and fuzzy logic inference. Firstly, the mathematical model is discussed to predict the physics-based GL （PGL）. Then, the diagnostic estimation model based on SVM is presented in detail to predict the empirical GL （EPL）. Thirdly, a fuzzy logic inference is adopted to fuse two GL predicted results. Finally, the GL prognostic model is verified by the run-to-failure data acquired from an accelerated life test of an aircraft bearing. The results show that the model provides a more practical and reliable prediction for the service life of bearings.

Grain Yield Prediction for Irrigation District Based on LS-SVM

Commonly used grain yield forecasting models were briefly reviewed, and a yield prediction model of irrigation district was established based on least squares support vector machines （LS-SVM）. The grain yield in irrigation district was analog calculated. And the test samples were used to compare with gray prediction, and neural network model. The maximum predicted error of least squares SVM was 7.12%, with an average error of 4.81%. The results showed that LS-SVM model has high prediction accuracy and strong generalization ability. So it could be used as a new method for irrigation district yield prediction

Support vector machines for emotion recognition in Chinese speech

Support vector machines (SVMs) are utilized for emotion recognition in Chinese speech in this paper. Both binary class discrimination and the multi class discrimination are discussed. It proves that the emotional features construct a nonlinear problem in the input space, and SVMs based on nonlinear mapping can solve it more effectively than other linear methods. Multi class classification based on SVMs with a soft decision function is constructed to classify the four emotion situations. Compared with principal component analysis (PCA) method and modified PCA method, SVMs perform the best result in multi class discrimination by using nonlinear kernel mapping.

Support vector machine for prediction of siRNA silencing efficacy

In order to assist the design of short interfering ribonucleic acids （siRNA）, 573 non-redundant siRNAs were collected from published literatures and the relationship between siRNAs sequences and RNA interference （RNAi） effect is analyzed by a support vector machine （SVM） based algorithm relied on a basebase correlation （BBC） feature. The results show that the proposed algorithm has the highest area under curve （AUC） value （0. 73） of the receive operating characteristic （ROC） curve and the greatest r value （0. 43） of the Pearson＇s correlation coefficient. This indicates that the proposed algorithm is better than the published algorithms on the collected datasets and that more attention should be paid to the base-base correlation information in future siRNA design.

Tribological properties and wear prediction model of TiC particles reinforced Ni-base alloy composite coatings

TiC particles reinforced Ni-based alloy composite coatings were prepared on 7005 aluminum alloy by plasma spray. The effects of load, speed and temperature on the tribological behavior and mechanisms of the composite coatings under dry friction were researched. The wear prediction model of the composite coatings was established based on the least square support vector machine (LS-SVM). The results show that the composite coatings exhibit smaller friction coefficients and wear losses than the Ni-based alloy coatings under different friction conditions. The predicting time of the LS-SVM model is only 12.93%of that of the BP-ANN model, and the predicting accuracies on friction coefficients and wear losses of the former are increased by 58.74%and 41.87%compared with the latter. The LS-SVM model can effectively predict the tribological behavior of the TiCP/Ni-base alloy composite coatings under dry friction.

ESSENTIAL RELATIONSHIP BETWEEN DOMAIN-BASED ONE-CLASS CLASSIFIERS AND DENSITY ESTIMATION

One-class support vector machine （OCSVM） and support vector data description （SVDD） are two main domain-based one-class （kernel） classifiers. To reveal their relationship with density estimation in the case of the Gaussian kernel, OCSVM and SVDD are firstly unified into the framework of kernel density estimation, and the essential relationship between them is explicitly revealed. Then the result proves that the density estimation induced by OCSVM or SVDD is in agreement with the true density. Meanwhile, it can also reduce the integrated squared error （ISE）. Finally, experiments on several simulated datasets verify the revealed relationships.

Phishing detection method based on URL features

In order to effectively detect malicious phishing behaviors, a phishing detection method based on the uniform resource locator （URL） features is proposed. First, the method compares the phishing URLs with legal ones to extract the features of phishing URLs. Then a machine learning algorithm is applied to obtain the URL classification model from the sample data set training. In order to adapt to the change of a phishing URL, the classification model should be constantly updated according to the new samples. So, an incremental learning algorithm based on the feedback of the original sample data set is designed. The experiments verify that the combination of the URL features extracted in this paper and the support vector machine （SVM） classification algorithm can achieve a high phishing detection accuracy, and the incremental learning algorithm is also effective.

Acoustic emission source identification based on harmonic wavelet packet and support vector machine

In order to solve the fatigue damage identification problem of helicopter moving components, a new approach for acoustic emission （AE） source type identification based on the harmonic wavelet packet （HWPT） feature extraction and the hierarchy support vector machine （H-SVM） classifier is proposed. After a four-level decomposition of the HWPT, the energy feature of AE signals in different frequency bands is extracted, which overcomes the shortcomings of the traditional wavelet packet including energy leakage, and inflexible frequency band selection and different frequency resolutions on different levels. The H-SVM classifier is trained with a subset of the experimental data for known AE source types and tested using the remaining set of data. The results of pressure-off experiments on the specimens of carbon fiber materials indicate that the proposed approach can effectively implement the AE source type identification, and has a better performance in terms of computational efficiency and identification accuracy than the wavelet packet （WPT） feature extraction.

Combination Computing of Support Vector Machine, Support Vector Regression and Molecular Docking for Potential Cytochrome P450 1A2 Inhibitors

The computational approaches of support vector machine （SVM）, support vector regression （SVR） and molecular docking were widely utilized for the computation of active compounds. In this work, to improve the accuracy and reliability of prediction, the strategy of combining the above three computational approaches was applied to predict potential cytochrome P450 1A2 （CYP1A2） inhibitors. The accuracy of the optimal SVM qualitative model was 99.432%, 97.727%, and 91.667% for training set, internal test set and external test set, respectively, showing this model had high discrimination ability. The R2 and mean square error for the optimal SVR quantitative model were 0.763, 0.013 for training set, and 0.753, 0.056 for test set respectively, indicating that this SVR model has high predictive ability for the biolog-ical activities of compounds. According to the results of the SVM and SVR models, some types of descriptors were identi ed to be essential to bioactivity prediction of compounds, including the connectivity indices, constitutional descriptors and functional group counts. Moreover, molecular docking studies were used to reveal the binding poses and binding a n-ity of potential inhibitors interacting with CYP1A2. Wherein, the amino acids of THR124 and ASP320 could form key hydrogen bond interactions with active compounds. And the amino acids of ALA317 and GLY316 could form strong hydrophobic bond interactions with active compounds. The models obtained above were applied to discover potential CYP1A2 inhibitors from natural products, which could predict the CYPs-mediated drug-drug inter-actions and provide useful guidance and reference for rational drug combination therapy. A set of 20 potential CYP1A2 inhibitors were obtained. Part of the results was consistent with references, which further indicates the accuracy of these models and the reliability of this combinatorial computation strategy.

ONLINE PARSIMONIOUS LEAST SQUARES SUPPORT VECTOR REGRESSION AND ITS APPLICATION

A simple and effective mechanism is proposed to realize the parsimoniousness of the online least squares support vector regression （LS-SVR）, and the approach is called the OPLS-SVR for short. Hence, the response time is curtailed. Besides, an OPLS-SVR based analytical redundancy technique is presented to cope with the sensor failure and drift problems to guarantee that the provided signals for the aeroengine controller are correct and acceptable. Experiments on the sensor failure and drift show the effectiveness and the validity of the proposed analytical redundancy.

Application of support vector machine in trip chaining pattern recognition and analysis of explanatory variable effects

In order to improve the accuracy of travel demand forecast and considering the distribution of travel behaviors within time dimension, a trip chaining pattern recognition model was established based on activity purposes by applying three methods： the support vector machine （SVM） model, the radial basis function neural network （RBFNN） model and the multinomial logit （MNL） model. The effect of explanatory factors on trip chaining behaviors and their contribution to model performace were investigated by sensitivity analysis. Results show that the SVM model has a better performance than the RBFNN model and the MNL model due to its higher overall and partial accuracy, indicating its recognition advantage under a smai sample size scenario. It is also proved that the SVM model is capable of estimating the effect of multi-category factors on trip chaining behaviors more accurately. The different contribution of explanatory, factors to trip chaining pattern recognition reflects the importance of refining trip chaining patterns ad exploring factors that are specific to each pattern. It is shown that the SVM technology in travel demand forecast modeling and analysis of explanatory variable effects is practical.

NOVEL WEIGHTED LEAST SQUARES SUPPORT VECTOR REGRESSION FOR THRUST ESTIMATION ON PERFORMANCE DETERIORATION OF AERO-ENGINE

A thrust estimator with high precision and excellent real-time performance is needed to mitigate perfor- mance deterioration for future aero-engines. A weight least squares support vector regression is proposed using a novel weighting strategy. Then a thrust estimator based on the proposed regression is designed for the perfor- mance deterioration. Compared with the existing weighting strategy, the novel one not only satisfies the require- ment of precision but also enhances the real-time performance. Finally, numerical experiments demonstrate the effectiveness and feasibility of the proposed weighted least squares support vector regression for thrust estimator. Key words ： intelligent engine control; least squares ; support vector machine ; performance deterioration

Application of a compound controller based on fuzzy control and support vector machine to ship's boiler-turbine coordinated control system

Multivariables, strong coupling, nonlinearity, and large delays characterize the boiler-turbine coordinated control systems for ship power equipment. To better deal with these conditions, a compound control strategy based on a support vector machine （SVM） with inverse identification was proposed and applied to research simulating coordinated control systems. This method combines SVM inverse control and fuzzy control, taking advantage of the merits of SVM inverse controls which can be designed easily and have high reliability, and those of fuzzy controls, which respond rapidly and have good anti-jamming capability and robustness. It ensures the controller can be controlled with near instantaneous adjustments to maintain a steady state, even if the SVM is not trained well. The simulation results show that the control quality of this fuzzy-SVM compound control algorithm is high, with good performance in dynamic response speed, static stability, restraint of overshoot, and robustness.

BOOSTING SPARSE LEAST SQUARES SUPPORT VECTOR REGRESSION (BSLSSVR) AND ITS APPLICATION TO THRUST ESTIMATION

In order to realize direct thrust control instead of conventional sensors-based control for aero-engine, a thrust estimator with high accuracy is designed by using the boosting technique to improve the performance of least squares support vector regression （LSSVR）. There exist two distinct features compared with the conven- tional boosting technique： （1） Sampling without replacement is used to avoid numerical instability for modeling LSSVR. （2） To realize the sparseness of LSSVR and reduce the computational complexity, only a subset of the training samples is used to construct LSSVR. Thus, this boosting method for LSSVR is called the boosting sparse LSSVR （BSLSSVR）. Finally, simulation results show that BSLSSVR-based thrust estimator can satisfy the requirement of direct thrust control, i.e. , maximum absolute value of relative error of thrust estimation is not more than 5‰.

Feature Extraction of Stored-grain Insects Based on Ant Colony Optimization and Support Vector Machine Algorithm

[Objective] The aim was to study the feature extraction of stored-grain insects based on ant colony optimization and support vector machine algorithm, and to explore the feasibility of the feature extraction of stored-grain insects. [Method] Through the analysis of feature extraction in the image recognition of the stored-grain insects, the recognition accuracy of the cross-validation training model in support vector machine （SVM） algorithm was taken as an important factor of the evaluation principle of feature extraction of stored-grain insects. The ant colony optimization （ACO） algorithm was applied to the automatic feature extraction of stored-grain insects. [Result] The algorithm extracted the optimal feature subspace of seven features from the 17 morphological features, including area and perimeter. The ninety image samples of the stored-grain insects were automatically recognized by the optimized SVM classifier, and the recognition accuracy was over 95%. [Conclusion] The experiment shows that the application of ant colony optimization to the feature extraction of grain insects is practical and feasible.

Semi-supervised least squares support vector machine algorithm:application to offshore oil reservoir

At the early stages of deep-water oil exploration and development, fewer and further apart wells are drilled than in onshore oilfields. Supervised least squares support vector machine algorithms are used to predict the reservoir parameters but the prediction accuracy is low. We combined the least squares support vector machine （LSSVM） algorithm with semi-supervised learning and established a semi-supervised regression model, which we call the semi-supervised least squares support vector machine （SLSSVM） model. The iterative matrix inversion is also introduced to improve the training ability and training time of the model. We use the UCI data to test the generalization of a semi-supervised and a supervised LSSVM models. The test results suggest that the generalization performance of the LSSVM model greatly improves and with decreasing training samples the generalization performance is better. Moreover, for small-sample models, the SLSSVM method has higher precision than the semi-supervised K-nearest neighbor （SKNN） method. The new semi- supervised LSSVM algorithm was used to predict the distribution of porosity and sandstone in the Jingzhou study area.

SVM method for predicting the thickness of sandstone

The Support Vector Machine （SVM） method can be used to set up a nonlinear function prediction model. It is based on the small sample learning theory. The kernel function can be constructed automatically based on the actual sample data by using the SVM method. As a result, the function not only gets a higher fit precision but is also better generalized. The frequency spectrum and seismic waveform are related by Fourier transform, so they are two different forms of the same physical phenomenon. The variety of waveform character reflects stratigraphic differences and frequency spectrum differences reflect the variation of lithology, fluid composition, and formation thickness. It directly predicts sandstone thickness using the seismic waveform. This not only fully utilizes the seismic information but also greatly increases the accuracy of the prediction. Model examples and actual applications show the applicability of this method.