Earticle

개인신용평가는 은행이 대출을 승인할 때 수익성 있는 의사결정을 적절히 유도할 수 있는 효과적인 도구이다. 최근 많은 분류 알고리즘 및 모델이 개인신용평가에 사용되고 있다. 개인신용평가 기법은 대체로 통계적 방법과 비 통계 적 방법으로 구분된다. 통계적 방법에는 선형회귀분석, 판별분석, 로지스틱 회귀분석, 의사결정나무 등이 포함된다. 비 통 계적 방법에는 선형계획법, 신경망, 유전자 알고리즘 및 Support Vector Machines 등이 포함된다. 그러나 신용평가모형 개발을 위해 어떠한 방법이 최선인지에 관해서는 일관된 결론을 내리기는 어렵다. 본 논문에서는 중국 금융기관의 개인 신용 데이터를 사용하여 가장 대표적인 신용평가 기법인 로지스틱 회귀분석, 신경망 그리고 Support Vector Machines의 성능을 비교하고자 한다. 구체적으로, 세 가지 모형을 각각 구축하여 고객을 분류하고 분석 결과를 비교하였다. 분석결과 에 따르면, Support Vector Machines이 로지스틱 회귀분석과 신경망보다 더 나은 성능을 가지는 것으로 나타났다.

Personal credit scoring is an effective tool for banks to properly guide decision profitably on granting loans. Recently, many classification algorithms and models are used in personal credit scoring. Personal credit scoring technology is usually divided into statistical method and non-statistical method. Statistical method includes linear regression, discriminate analysis, logistic regression, and decision tree, etc. Non-statistical method includes linear programming, neural network, genetic algorithm and support vector machine, etc. But for the development of the credit scoring model, there is no consistent conclusion to be drawn regarding which method is the best. In this paper, we will compare the performance of the most common scoring techniques such as logistic regression, neural network, and support vector machines using personal credit data of the financial institution in China. Specifically, we build three models respectively, classify the customers and compare analysis results. According to the results, support vector machine has better performance than logistic regression and neural networks.

Corporate credit-rating prediction using statistical and artificial intelligence (AI) techniques has received considerable research attention in the literature. In recent years, multiclass support vector machines (MSVMs) have become a very appealing machine-learning approach due to their good performance. Until now, researchers have proposed a variety of techniques for adapting support vector machines (SVMs) to multiclass classification, since SVMs were originally devised for binary classification. However, the studies that applied MSVMs in predicting bond rating just adopted a few techniques. This study is designed to evaluate all MSVM techniques proposed in the literature for the prediction of corporate bond rating in Korea. To do this, we applied six different techniques of MSVMs, and compared the prediction performances. We also examined some modified version of conventional MSVM techniques.

Support vector machines (SVMs), a machine learning technique, has been applied to not only binary classification problems such as bankruptcy prediction but also multi-class problems such as corporate credit ratings. However, in general, the performance of SVMs can be easily worse than the best alternative model to SVMs according to the selection of predictors, even though SVMs has the distinguishing feature of successfully classifying and predicting in a lot of dichotomous or multi-class problems. For overcoming the weakness of SVMs, this study has proposed an approach for selecting features for multi-class SVMs that utilize the impurity measures of classification trees. For the selection of the input features, we employed the C4.5 and CART algorithms, including the stepwise method of discriminant analysis, which is a well-known method for selecting features. We have built a multi-class SVMs model for credit rating using the above method and presented experimental results with data regarding S&P 500 companies.

Prediction of corporate bankruptcies has long been an important topic and has been studied extensively in the finance and management literature because it is an essential basis for the risk management of financial institutions. Recently, support vector machines (SVMs) are becoming popular as a tool for bankruptcy prediction because they use a risk function consisting of the empirical error and a regularized term which is derived from the structural risk minimization principle. In addition, they don't require huge training samples and have little possibility of overfitting. However. in order to Use SVM, a user should determine several factors such as the parameters ofa kernel function, appropriate feature subset, and proper instance subset by heuristics, which hinders accurate prediction results when using SVM In this study, we propose a novel hybrid SVM classifier with simultaneous optimization of feature subsets, instance subsets, and kernel parameters. This study introduces genetic algorithms (GAs) to optimize the feature selection, instance selection, and kernel parameters simultaneously. Our study applies the proposed model to the real-world case for bankruptcy prediction. Experimental results show that the prediction accuracy of conventional SVM may be improved significantly by using our model.

Conventional Support Vector Machines (SVMs) have been utilized as classifiers for binary classification problems. However, certain real world problems, including corporate bond rating, cannot be addressed by binary classifiers because these are multi-class problems. For this reason, numerous studies have attempted to transform the original SVM into a multiclass classifier. These studies, however, have only considered nominal classification problems. Thus, these approaches have been limited by the existence of multiclass classification problems where classes are not nominal but ordinal in real world, such as corporate bond rating and multiclass customer classification. In this study, we adopt a novel multiclass SVM which can address ordinal classification problems using ordinal pairwise partitioning (OPP). The proposed model in our study may use fewer classifiers, but it classifies more accurately because it considers the characteristics of the order of the classes. Although it can be applied to all kinds of ordinal multiclass classification problems, most prior studies have applied it to finance area like bond rating. Thus, this study applies it to a real world customer level classification case for implementing customer relationship management. The result shows that the ordinal multiclass SVM model may also be effective for customer level classification.

Corporate credit rating is a very important factor in the market for corporate debt. Information concerning corporate operations is often disseminated to market participants through the changes in credit ratings that are published by professional rating agencies, such as Standard and Poor’s (S&P) and Moody’s Investor Service. Since these agencies generally require a large fee for the service, and the periodically provided ratings sometimes do not reflect the default risk of the company at the time, it may be advantageous for bond-market participants to be able to classify credit ratings before the agencies actually publish them. As a result, it is very important for companies (especially, financial companies) to develop a proper model of credit rating.From a technical perspective, the credit rating constitutes a typical, multiclass, classification problem be-cause rating agencies generally have ten or more categories of ratings. For example, S&P’s ratings range from AAA for the highest-quality bonds to D for the lowest-quality bonds. The professional rating agencies emphasize the importance of analysts’subjective judgments in the determination of credit ratings. However, in practice, a mathematical model that uses the financial variables of companies plays an important role in determining credit ratings, since it is convenient to apply and cost efficient. . These financial variables include the ratios that represent a company’s leverage status, liquidity status, and profitability status.Several statistical and artificial intelligence (AI) techniques have been applied as tools for predicting credit ratings. Among them, artificial neural networks are most prevalent in the area of finance because of their broad applicability to many business problems and their preeminent ability to adapt. However, artificial neural networks also have many defects, including the difficulty in determining the values of the control parameters and the number of processing elements in the layer as well as the risk of over-fitting. Of late, because of their robustness and high accuracy, support vector machines (SVMs) have become popular as a solution for problems with generating accurate prediction. An SVM’s solution may be globally optimal because SVMs seek to minimize structural risk. On the other hand, artificial neural network models may tend to find locally optimal solutions because they seek to minimize empirical risk. In addition, no parameters need to be tuned in SVMs, barring the upper bound for non-separable cases in linear SVMs. Since SVMs were originally devised for binary classification, however they are not intrinsically geared for multiclass classifications as in credit ratings. Thus, researchers have tried to extend the original SVM to multiclass classification.Hitherto, a variety of techni-ques to extend standard SVMs to multiclass SVMs (MSVMs) has been proposed in the literature Only a few types of MSVM are, however, tested using prior studies that applyMSVMs to credit ratings studies..In this study, we examined six different techniques of MSVMs: (1) One-Against-One, (2) One-Against-All, (3) DAGSVM, (4) ECOC, (5) Method of Weston and Watkins, and (6) Method of Crammer and Singer. In addition, we examined the prediction accuracy of some modified version of conventional MSVM techniques. To find the most appropriate technique of MSVMs for corporate bond rating, we applied all the techniques of MSVMs to a real-world case of credit rating in Korea. The best application is in corporate bond rating, which is the most frequently studied area of credit rating for specific debt issues or other financial obligations. For our study the research data were collected from National Information and Credit Evaluation, Inc., a major bond-rating company in Korea. The data set is comprised of the bond-ratings for the year 2002 and various financial variables for 1,295 companies from the manufacturing industry in Korea.We compared the results of these techniques with one another, and with those of traditional methods for credit ratings, such as multiple discriminant analysis (MDA), multinomial logistic regression (MLOGIT), and artificial neural networks (ANNs). As a result, we found that DAGSVM with an ordered list was the best approach for the prediction of bond rating. In addition, we found that the modified version of ECOC approach can yield higher prediction accuracy for the cases showing clear patterns.

본 논문에서는 서포트 벡터 머신(Support Vector Machine: SVM) 기법을 사용하여 드론과 조류의 움직임을 고려하여 생성된 마이크로 도플러 신호를 분석하여 드론과 조류를 구별하는 시뮬레이션 연구를 제시하였다. 두 개체의 스펙트로그램에서 얻은 시간-주파수 특징은 분류를 위해 평균 스펙트럼 프로파일로 변환된다. SVM 분류기는 신호의 스펙 트로그램을 활용하여 학습되며, 드론과 조류의 동작 매개변수(드론 속도, 드론 블레이드 회전, 조류 날갯짓)의 변동성을 종합적으로 고려하여 분석되었다. 제안된 방법은 드론 블레이드 회전 속도 2,000rpm(±25% 잡음)과 본체 속도 8m/s(±50% 잡음)를 기반으로 한 학습 신호를 사용하여 블레이드 회전 속도가 1,600rpm 정도로 낮은 경우에도 거의 100%의 분류 정확도를 보였다. 블레이드 회전으로 인한 도플러 주파수가 본체 이동에 의한 도플러 주파수보다 상당히 높기 때문에 본체 속도가 분류 성능에 미치는 영향은 무시할 수 있다. 본 연구 결과는 소형 드론을 구별하는 데 있어 SVM 기반 분류 기술의 효과를 제시하였으며, 이 방법의 계산 효율성 측면에서 실시간 감시 시나리오에 매우 적합할 것으로 판단된다.

In this paper, a simulation study for distinguishing between drones and birds by analyzing simulated radar micro-Doppler signatures using a Support Vector Machine (SVM) classifier is presented. Time-frequency features obtained from spectrograms are condensed into average spectral profiles for classification purposes. The SVM classifier is trained using these spectrogram-based features, incorporating variability in drone and bird motion parameters to improve its generalization capability. Results show that with training signals based on a drone rotor speed of 2,000rpm(±25% noise) and body speed of 8m/s(±50% noise), the classifier achieved nearly 100% accuracy for test cases with rotor speeds as low as 1,600rpm. Since the Doppler frequency generated by wing rotation is significantly higher than that produced by body translation, the effect of body velocity on classification performance is negligible. The results confirm the effectiveness of the SVM-based technique in small drone discrimination, and its computational efficiency makes it well-suited for real-time surveillance scenarios.

Any economic and social activities are inseparable from a specific area, and in different regions often mean different development policies. China attaches great importance to the division of the region to a certain area in order to develop targeted policies. Currently, all provinces in China are divided into eastern, central, and western regions, 11 provinces in eastern region, 12 provinces in western region, and 8 provinces in central region. Since the founding, especially since the reform and opening up, China’s regional policy has undergone so many alternates between comprehensive development and focusing development, each policy adjustments have had a significant impact on the development of the relation region. Compared to the current administrative division, this article uses clustering methods to classify China provinces, and support vector machine for testing. The study found that three regions administrative division of China currently used cannot adapt to the actual development of the current situation, we must make the appropriate adjustments in order to provide a basis for the development of more targeted regional policy.

任何经济社会活动都离不开特定的区域, 处于不同的区域往往意味着不同的 发展政策。为增强宏观调控的有效性, 在中国的经济社会发展中非常重视将某一 个地区划分到一类区域当中, 从而对该区域制定有针对性的政策。目前, 中国 将所有省份分为东部、中部、西部三个区域, 东部11个省份, 西部12个省份, 中 部8个省份。建国以来, 特别是改革开放以来, 中国的区域政策经历了重点发展、 全面发展、重点发展与全面发展相结合等多次政策交替变换, 每次政策调整均对 相应区域的发展产生了重大影响。本文使用聚类方法对中国省份进行了分类, 与目前东中西行政区域划分进行对比分析, 并使用支持向量机进行检验。研究发 现, 中国目前采用的东中西行政区域划分已经不能适应目前的实际发展状况, 必 须进行相应的调整, 以便为制定更有针对性的区域政策提供参考。

Remote protein homology detection has been widely used as a part of the analysis of protein structure and function. In this study, the good quality of protein feature vectors is the main aspect to detect remote protein homology; as it will assist discriminative classifier model to discriminate all the proteins into homologue or non-homologue members precisely. In order for the protein feature vectors to be characterized as having good quality, the feature vectors must contain high protein structural similarity information and are represented in low dimension which is free from any contaminated data. In this study, the contaminated data which originates from protein dataset was investigated. This contaminated data may prevent remote protein homology detection framework to produce the best representation of high protein structural similarity information in order to detect the homology of proteins. To reduce the contaminated data and extract high protein structural similarity information, some research has been done on the extraction of protein feature vectors and protein similarity. The extraction of protein feature vectors of good quality is believed could assist in getting better result for remote protein homology detection. Where, the good quality of protein feature vectors containing the useful protein similarity information and represent in low dimension will be used to identify protein family precisely by discriminative classifier model. Referring to this factor, a method which combines Protein Substring Scoring (PSS) and Pairwise Protein Substring Alignment (PPSA) from sequence comparison model, chi-square and Singular Value Decomposition (SVD) from generative model, and Support Vector Machine (SVM) as discriminative classifier model is introduced.

Remote protein homology detection is a problem of detecting evolutionary relationship between proteins at low sequence similarity level. Among several problems in remote protein homology detection include the questions of determining which combination of multiple alignment and classification techniques is the best as well as the misalignment of protein sequences during the alignment process. Therefore, this paper deals with remote protein homology detection via assessing the impact of using structural information on protein multiple alignments over sequence information. This paper further presents the best combinations of multiple alignment and classification programs to be chosen. This paper also improves the quality of the multiple alignments via integration of a refinement algorithm. The framework of this paper began with datasets preparation on datasets from SCOP version 1.73, followed by multiple alignments of the protein sequences using CLUSTALW, MAFFT, ProbCons and T-Coffee for sequence-based multiple alignments and 3DCoffee, MAMMOTH-mult, MUSTANG and PROMALS3D for structural-based multiple alignments. Next, a refinement algorithm was applied on the protein sequences to reduce misalignments. Lastly, the aligned protein sequences were classified using the pHMMs generative classifier such as HMMER and SAM and also SVMs discriminative classifier such as SVM-Fold and SVM-Struct. The performances of assessed programs were evaluated using ROC, Precision and Recall tests. The result from this paper shows that the combination of refined SVM-Struct and PROMALS3D performs the best against other programs, which suggests that this combination is the best for RPHD. This paper also shows that the use of the refinement algorithm increases the performance of the multiple alignments programs by at least 4%.

This paper uses Support Vector Machines (SVM) to fuse multiple classifiers for an offline signature system. From the signature images, global and local features are extracted and the signatures are verified with the help of Gaussian empirical rule, Euclidean and Mahalanobis distance based classifiers. SVM is used to fuse matching scores of these matchers. Finally, recognition of query signatures is done by comparing it with all signatures of the database. The proposed system is tested on a signature database contains 5400 offline signatures of 600 individuals and the results are found to be promising.

Speed control of hydraulic actuator is mainly desirable in diverse hydraulic systems including those for agriculture and most of them are for manufacturing. In all these systems the main demand is to accurately control the speed of the hydraulic actuator under variable operating conditions including various set-points and disturbances. In this paper a generalized predictive control (GPC) which is well known is used based on least squares support vector machines (LS-SVM) as a model reference control. The main drawback of LS-SVM is that the sparseness of standard SVM is lost which means that some of the system parameters are equal to zero. In this contribution a new method for sparseness is introduced. The electro-hydraulic servo system is modeled recursively each sample based on LS-SVM using linear kernel function avoiding linearization when radial basis function (RBF) is employed. Based on the model obtained the control signal is calculated at each sample using GPC. The presented controller performance is compared to that of self-tuning PID controller based on recursive least squares (RLS) algorithm.

Considering that intrusion detection systems and anomaly clearly recognize malicious activity. Nowadays, data mining based intrusion detection systems, security and more rapidly detect attacks.Therefore,in this article we use a combination of k-means clustering algorithm and is used supervised support vector machine algorithm to find the best line separator. This is leading to the separation of normal and attack traffic.

Terror attacks are increased worldwide. The early detection of weapons is an important objective for security specialists. In this paper, we proposed an automated surveillance system for detecting fire weapons in cluttered scene. First SIFT features are extracted from the collection of images. Second, K-means clustering is adopted for clustering the SIFT features. Third, a word vocabulary based histogram is implemented by counting occurrences of the extracted clusters in each image. The histogram is the input to Support Vector Machine that will be trained on the collection of images. Finally, the trained SVM is the system classifier that will decide if new image contains a weapon or not. The main contributions of the paper is to adopt the visual words classification scheme in detecting fire weapons. In addition, we used RANSAC to reduce the matching outliers. The system showed high accuracy in detecting fire weapons in images and video surveillance systems.

The basic task in intrusion detection system is to classify network activities as normal or abnormal while minimizing misclassification. In literature, various machine learning and data mining techniques have been applied to Intrusion Detection Systems (IDSs) to protect the special computer systems, vulnerable traffics cyber-attacks for computer networks. In addition, Support Vector Machine (SVM) is applied as the classification techniques in literature. However, there is a lack of review for the IDS method using SVM as the classifier. The objective of this paper is to review the contemporary literature and to provide a critical evaluation of various techniques of intrusion detection using SVM as classifier. We analyze and identify the strengths and limitations of various SVM usages as classifier in IDS systems. This paper also highlights the usefulness of SVM in IDS system for network security environment with future direction.

Indoor localization based on existing WiFi signal strength is becoming increasingly prevalent and ubiquitous. The user-based localization algorithm utilizes the information of the Received Signal Strength(RSS) from the surrounding access points(APs) to determine the user position. In this paper, focusing on the development of a user localization uses existing WiFi environment for its low cost and ease of deployment. We propose an indoor localization of WiFi based on support vector machines(ILW-SVM), and use the bilinear median interpolation method(BMIM) to reduce the calibration effort on creating fingerprint map while still retaining the accuracy of user localization. According to comparison of accuracy of three different kernel functions, choosing the radial basis function(RBF) as kernel function. In addition, we also propose improved ILW-SVM algorithm to solve the indoor localization that nearest neighbor points are not concentrated. At last, overall comparison of kNN, ILW-SVM and improved ILW-SVM in consideration of accuracy. Experimental results indicate that the proposed algorithm can effectively reduce the calibration effort and exhibit superior performance in terms of localization accuracy and stabilization.

Due to increase in industrial and anthropogenic activities, air pollution has been a serious environmental problem all over the world. It was found that harmful emission into the air is a symbol for environmental force that affects seriously man’s health, natural life and agriculture; thus leading to major loss of the nation’s economy. In this paper, the prediction of the surface ozone layer problem is explored. A comparison between two types of Artificial Neural Networks (ANN) (i.e. back propagation and Radial Basis Functions (RBF) networks) and the Support Vector Machines (SVM) techniques for short prediction of surface ozone is conclusively demonstrated. Three models which predict the expected values of the surface ozone based on three variables (i.e. Nitrogen-di-oxide, temperature and Relative Humidity) will be presented.

The Attitude control model for AUV is highly nonlinear and strong coupling six degrees of freedom model. With it, and complex disturbance of the marine environment a new control strategy chaos least squares support vector machines is proposed. The control model is infinitely approximated by least squares support vector machine. The parameters are optimized Chaos theory. It is formed the Chaos-LSSVM model. The purpose is to obtain excellent attitude control of quality. The Chaos-LSSVM predictive controller for AUV attitude is designed combined with a simplified longitudinal hydrodynamic equations and the Chaos-LSSVM model. Meanwhile the stability of the attitude control system is ensured by the Lyapunov stability theory. Finally, examples of simulation results show that the control model can obtain a good control effect, improve the underwater vehicle dynamic positioning attitude control effectiveness and robustness.