scholarly journals National Bank of Ukraine Econometric Model for the Assessment of Banks’ Credit Risk and Support Vector Machine Alternative

2015 ◽  
pp. 52-72
Author(s):  
Dmytro Pokidin
Keyword(s):  
Neural Networks ◽  
Support Vector Machine ◽  
Logit Model ◽  
Econometric Model ◽  
Credit Scoring ◽  
Econometric Models ◽  
Support Vector ◽  
Practical Implementation ◽  
Default Prediction ◽  
National Bank

Econometric models of credit scoring started with the introduction of Altman’s simple z-model in 1968, but since then these models have become more and more sophisticated, some even use Artificial Neural Networks (ANN) and Support Vector Machine (SVM) techniques. This paper focuses on the use of SVM as a model for default prediction. I start with an introduction to SVM as well as to some of its widespread alternatives. Then, these different techniques are used to model NBU data on banks’ clients, which allows us to compare the accuracy of SVM to the accuracy of other models. While SVM is generally more accurate, I discuss some of the features of SVM that make its practical implementation controversial. I then discuss some ways for overcoming those features. I also present the results of the Logistic Regression (Logit) model which will be used by the NBU.

Sensor Dynamic Modeling Based on LS-SVM and NGA

2008 ◽  
Vol 381-382 ◽  
pp. 439-442
Author(s):  
Qi Wang ◽  
Zhi Gang Feng ◽  
K. Shida
Keyword(s):  
Genetic Algorithm ◽  
Neural Networks ◽  
Support Vector Machine ◽  
Least Squares ◽  
Dynamic Model ◽  
Dynamic Modeling ◽  
Support Vector ◽  
Local Minima ◽  
Highly Nonlinear ◽  
Sharing Function

Least squares support vector machine (LS-SVM) combined with niche genetic algorithm (NGA) are proposed for nonlinear sensor dynamic modeling. Compared with neural networks, the LS-SVM can overcome the shortcomings of local minima and over fitting, and has higher generalization performance. The sharing function based niche genetic algorithm is used to select the LS-SVM parameters automatically. The effectiveness and reliability of this method are demonstrated in two examples. The results show that this approach can escape from the blindness of man-made choice of LS-SVM parameters. It is still effective even if the sensor dynamic model is highly nonlinear.

IMPROVEMENT OF DOSE ESTIMATION PROCESS USING ARTIFICIAL NEURAL NETWORKS

2018 ◽  
Vol 184 (1) ◽  
pp. 36-43 ◽  
Author(s):  
Gal Amit ◽  
Hanan Datz
Keyword(s):  
Neural Networks ◽  
Support Vector Machine ◽  
Artificial Neural Networks ◽  
Support Vector ◽  
Physical Parameters ◽  
Machine Method ◽  
Algorithm Performance ◽  
Artificial Neural ◽  
First Time ◽  
Automatic Algorithm

Abstract We present here for the first time a fast and reliable automatic algorithm based on artificial neural networks for the anomaly detection of a thermoluminescence dosemeter (TLD) glow curves (GCs), and compare its performance with formerly developed support vector machine method. The GC shape of TLD depends on numerous physical parameters, which may significantly affect it. When integrated into a dosimetry laboratory, this automatic algorithm can classify ‘anomalous’ (having any kind of anomaly) GCs for manual review, and ‘regular’ (acceptable) GCs for automatic analysis. The new algorithm performance is then compared with two kinds of formerly developed support vector machine classifiers—regular and weighted ones—using three different metrics. Results show an impressive accuracy rate of 97% for TLD GCs that are correctly classified to either of the classes.

scholarly journals Adaptive control scheme based on the least squares support vector machine network

2011 ◽  
Vol 21 (4) ◽  
pp. 685-696 ◽  
Author(s):  
Tarek Mahmoud
Keyword(s):  
Neural Networks ◽  
Support Vector Machine ◽  
Adaptive Control ◽  
Least Squares ◽  
Nonlinear System Identification ◽  
Support Vector ◽  
Reference Trajectory ◽  
Control Scheme ◽  
Support Vector Machine Network ◽  
Adaptive Control Scheme

Adaptive control scheme based on the least squares support vector machine networkRecently, a new type of neural networks called Least Squares Support Vector Machines (LS-SVMs) has been receiving increasing attention in nonlinear system identification and control due to its generalization performance. This paper develops a stable adaptive control scheme using the LS-SVM network. The developed control scheme includes two parts: the identification part that uses a modified structure of LS-SVM neural networks called the multi-resolution wavelet least squares support vector machine network (MRWLS-SVM) as a predictor model, and the controller part that is developed to track a reference trajectory. By means of the Lyapunov stability criterion, stability analysis for the tracking errors is performed. Finally, simulation studies are performed to demonstrate the capability of the developed approach in controlling a pH process.

Sign in / Sign up

Export Citation Format

Share Document