Shian-Chang Huang

Evaluation of ANN and SVM classifiers as predictors to the diagnosis of students with learning disabilities

Due to the implicit characteristics of learning disabilities (LD), the identification or diagnosis of students with learning disabilities has long been a difficult issue. The LD diagnosis procedure usually involves interpreting some standard tests or checklist scores and comparing them to norms that are derived from statistical method. In this paper, we apply two well-known artificial intelligence techniques, artificial neural network (ANN) and support vector machine (SVM), to the LD diagnosis problem. To improve the overall identification accuracy, we also experiment with GA-based feature selection algorithms as the pre-processing step. To the best of our knowledge, this is the first attempt in applying ANN or SVM to similar application. The experimental results show that ANN in general performs better than SVM in this application, and the wrapper-based GA feature selection procedure can improve the LD identification accuracy, and among all, the combination of using SVM learner in the feature selection procedure and ANN learner in the classification stage results in feature set that achieves the best prediction accuracy. Most important of all, the study indicates that the ANN classifier can correctly identify up to 50% of the LD students with 100% confidence, which is much better than currently used LD diagnosis predictors derived through the statistical method. Consequently, a properly trained ANN classification model can be a strong predictor for use in the LD diagnosis procedure. Furthermore, a well-trained ANN model can also be used to verify whether a LD diagnosis procedure is adequate. In conclusion, we expect that AI techniques like ANN or SVM will certainly play an essential role in future LD diagnosis applications.

Chaos-based support vector regressions for exchange rate forecasting

This study implements a chaos-based model to predict the foreign exchange rates. In the first stage, the delay coordinate embedding is used to reconstruct the unobserved phase space (or state space) of the exchange rate dynamics. The phase space exhibits the inherent essential characteristic of the exchange rate and is suitable for financial modeling and forecasting. In the second stage, kernel predictors such as support vector machines (SVMs) are constructed for forecasting. Compared with traditional neural networks, pure SVMs or chaos-based neural network models, the proposed model performs best. The root-mean-squared forecasting errors are significantly reduced.

Integrating nonlinear graph based dimensionality reduction schemes with SVMs for credit rating forecasting

By integrating graph based nonlinear dimensionality reduction with support vector machines (SVMs), this study develops a novel prediction model for credit ratings forecasting. SVMs have been successfully applied in numerous areas, and have demonstrated excellent performance. However, due to the high dimensionality and nonlinear distribution of the input data, this study employed a kernel graph embedding (KGE) scheme to reduce the dimensionality of input data, and enhance the performance of SVM classifiers. Empirical results indicated that one-vs-one SVM with KGE outperforms other multi-class SVMs and traditional classifiers. Compared with other dimensionality reduction methods the performance improvement owing to KGE is significant.

Integrating recurrent SOM with wavelet-based kernel partial least square regressions for financial forecasting

This study implements a novel expert system for financial forecasting. In the first stage, wavelet analysis transforms the input space of raw data to a time-scale feature space suitable for financial forecasting, and then a Recurrent Self-Organizing Map (RSOM) algorithm is used for partitioning and storing temporal context of the feature space. In the second stage, multiple kernel partial least square regressors (as local models) that best fit partitioned regions are constructed for final forecasting. Compared with neural networks, pure SVMs or traditional GARCH models, the proposed model performs best. The root-mean-squared forecasting errors are significantly reduced.

Using Gaussian process based kernel classifiers for credit rating forecasting

The subprime mortgage crisis have triggered a significant economic decline over the world. Credit rating forecasting has been a critical issue in the global banking systems. The study trained a Gaussian process based multi-class classifier (GPC), a highly flexible probabilistic kernel machine, using variational Bayesian methods. GPC provides full predictive distributions and model selection simultaneously. During training process, the input features are automatically weighted by their relevances with respect to the output labels. Benefiting from the inherent feature scaling scheme, GPCs outperformed convectional multi-class classifiers and support vector machines (SVMs). In the second stage, conventional SVMs enhanced by feature selection and dimensionality reduction schemes were also compared with GPCs. Empirical results indicated that GPCs still performed the best.

Kernel local Fisher discriminant analysis based manifold-regularized SVM model for financial distress predictions

Support vector machines (SVM) have demonstrated excellent performance in numerous areas of pattern recognitions. However, traditional SVM does not make efficient use of both labeled training data and unlabeled testing data. Moreover, high dimensional and nonlinear distributed data generally degrade the performance of a classifier due to the curse of dimensionality in financial distress (or bankruptcy) predictions. To address these problems, this study proposes a novel hybrid classifier which integrates Kernel local Fisher discriminant analysis (KLFDA) with a manifold-regularized SVM (MR-SVM). KLFDA is employed to find an optimal projection which maximizes the margin between data points from different classes at each local area of data manifold, while MR-SVM data-dependently warps the structure of feature space to reflect the underlying geometry of the data manifold. Compared with other dimensionality reduction methods and conventional classifiers, the hybrid classifier performs best.

Rough Sets as a Knowledge Discovery and Classification Tool for the Diagnosis of Students with Learning Disabilities

Due to the implicit characteristics of learning disabilities (LDs), the diagnosis of students with learning disabilities has long been a difficult issue. Artificial intelligence techniques like artificial neural network (ANN) and support vector machine (SVM) have been applied to the LD diagnosis problem with satisfactory outcomes. However, special education teachers or professionals tend to be skeptical to these kinds of black-box predictors. In this study, we adopt the rough set theory (RST), which can not only perform as a classifier, but may also produce meaningful explanations or rules, to the LD diagnosis application. Our experiments indicate that the RST approach is competitive as a tool for feature selection, and it performs better in term of prediction accuracy than other rulebased algorithms such as decision tree and ripper algorithms. We also propose to mix samples collected from sources with different LD diagnosis procedure and criteria. By pre-processing these mixed samples with simple and rea...

Improving ANN classification accuracy for the identification of students with LDs through evolutionary computation

Due to the implicit characteristics of learning disabilities (LD), the identification and diagnosis of students with learning disabilities has long been a difficult issue. Identification of LD usually requires extensive man power and takes a long time. Our previous studies using smaller size samples have shown that ANN classifier can be a good predictor to the identification of students with learning disabilities. In this study, we focus on the genetic algorithm based feature selection method and evolutionary computation based parameters optimization algorithm on a larger size data set to explore the potential limit that ANN can achieve in this specific classification problem. In addition, a different procedure in combining evolutionary algorithms and neural networks is proposed. We use the back propagation method to train the neural networks and derive a good combination of parameters. Evolutionary algorithm is then used to search around the neighborhood of the previously acquired parameters to further improve the classification accuracy. The outcomes show that the procedure is effective, and with appropriate combinations of features and parameters setting, the accuracy of the ANN classifier can reach the 88.2% mark, the best we have achieved so far. The result again indicates that AI techniques do have their roles on the identification of students with learning disabilities.

A hybrid unscented kalman filter and support vector machine model in option price forecasting

This study develops a hybrid model that combines unscented Kalman filters (UKFs) and support vector machines (SVMs) to implement an online option price predictor. In the hybrid model, the UKF is used to infer latent variables and make a prediction based on the Black-Scholes formula, while the SVM is employed to capture the nonlinear residuals between the actual option prices and the UKF predictions. Taking option data traded in Taiwan Futures Exchange, this study examined the forecasting accuracy of the proposed model, and found that the new hybrid model is superior to pure SVM models or hybrid neural network models in terms of three types of options. This model can also help investors for reducing their risk in online trading.

Combining ICA with Kernel Based Regressions for Trading Support Systems on Financial Options

Options are highly non-linear and complicated products in financial markets. Owing to the high risk associated with option trading, investment on options is a knowledge-intensive industry. This study develops a novel decision support system for option trading. In the first stage, independent component analysis (ICA) is employed to uncover the independent hidden forces of the stock market that drive the price movement of an option. In the second stage, a dynamic kernel predictors are constructed for trading decisions. Comparing with convectional feature extractions and pure regression models, the performance improvement of the new method is significant and robust. The cumulated trading profits are substantialy increased. The resultant intelligent investment decision support system can help investors, fund managers and investment decision-makers make profitable decisions.