Tung-Kuang Wu

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.

A distributed and cooperative black hole node detection and elimination mechanism for ad hoc networks

A mobile node in ad hoc networks may move arbitrarily and acts as a router and a host simultaneously. Such a characteristic makes nodes in MANET vulnerable to potential attacks. The black hole problem, in which some malicious nodes pretend to be intermediate nodes of a route to some given destinations and drop any packet that subsequently goes through it, is one of the major types of attack. In this paper, we propose a distributed and cooperative mechanism to tackle the black hole problem. The mechanism is distributed so that it can fit with the ad hoc nature of network, and nodes in the protocol work cooperatively together so that they can analyze, detect, and eliminate possible multiple black hole nodes in a more reliable fashion. Simulation results show that our method achieves a high black hole detection rate and good packet delivery ratio, while the overhead is comparatively lower as the network traffic increases.

A low overhead dynamic route repairing mechanism for mobile ad hoc networks

Ad hoc networks are wireless networks with no fixed infrastructure. Each mobile node in the network functions as a router that discovers and maintains routes for other nodes. These nodes may move arbitrarily, therefore network topology changes frequently and unpredictably. Many routing protocols have been designed for ad hoc networks. However, most of these kinds of protocols are not able to react fast enough to maintain routing. In this paper, we propose a new protocol that repairs the broken route by using information provided by nodes overhearing the main route communication. When links go down, our protocol intelligently replaces these failed links or nodes with backup ones that are adjacent to the main route. Theoretical analysis reveals that, in a given circumstance, our proposed protocol can find a backup route in more than 60% of time. Simulation results also demonstrate that our protocol achieves better (or as good) in terms of the packet delivery rate, control packet overhead and communication delay than the major ad hoc routing protocols under light and moderate traffic conditions.

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.

A highly topology adaptable ad hoc routing protocol with complementary preemptive link breaking avoidance and path shortening mechanisms

Ad-hoc on-demand distance vector routing (AODV) is a well-known routing protocol for mobile ad hoc networks. The original AODV protocol works in a semi-dynamic fashion, by establishing a route on demand and using that route until it breaks. However, to suit the changing network topology of ad hoc networks, more aggressive and adaptable routing strategies are required. A number of researches have proposed improving AODV performance by locally repairing broken links, predicting and replacing potentially vulnerable links, or shortening a link through removing redundant nodes from the transmission path. Although local repair may relieve some problems, it usually results in longer paths and thus a considerable performance drop in heavy traffic conditions. There are also issues regarding packet loss and communication delay due to route rebuilding once the link is broken. Predicting and replacing potentially vulnerable links may require special hardware, additional tables to maintain, or other extra overhead. Finally, path shortening may result in shorter and more efficient routes, but there is no guarantee that the new paths will be robust. This paper proposes integrating preemptive link breaking avoidance and path shortening mechanisms into a modified AODV protocol. However, the difficult issue lies in determining the right timing to initiate the two independent mechanisms so that the two dynamically and complementarily operating mechanisms can work together to improve the routing performance. Through numerical analysis and simulation, we have arranged a simple parameter setting for controlling the activation of each mechanism at the appropriate time. The proposed combination is a highly dynamic ad hoc routing protocol that is capable of adapting itself to the changing network topology and achieving extremely good performance in various routing performance metrics. Extensive simulations show that each of the two schemes alone improves AODV performance. More importantly, the integrated protocol performs even better in terms of data delivery rate, average delay time, and network overhead. To be more specific, in the best cases our protocol can reduce up to 82% in control overhead and 66% in delay time, while achieving 12% more in data delivery rate comparing to AODV.

A Novel Approach to the Fixed Channel Assignment Problem

A critical task in the design of a cellular radio network is to determine a spectrum-efficient and conflict-free allocation of channels among the cells. In this paper, we propose a new approach to the Fixed Channel Assignment (FCA) problem. By preserving the co-site channel constraint throughout our algorithm and adopting a fine-tuning procedure to escape from a local minimum, we reduce the overall execution time and improve the convergence rate. Simulation results show that our algorithm achieves a very high rate of convergence to solutions for eight benchmark problems. Furthermore, the number of iterations our algorithm requires is fewer than previous results.

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...

Localized Routing Protocols Based on Minimum Balanced Tree in Wireless Sensor Networks

Reducing energy consumption and prolonging lifetime of network to reduce the amount of packet loss are important issues in wireless sensor networks. Many researches derive the minimum hop path for each sensor to transmit its corresponding data to the sink. The sensors in the path forward the data. However, some common sensors in many forwarding paths will consume much more energy, and then they will die soon. Besides, the establishment and maintenance of the above routing need the whole information of the network, and this will consume more energy in gathering and synchronizing the locations of all sensors. In this paper, each sensor using the information of neighboring sensors derives the minimum hop path to the sink, and with the knowledge of the loading of its up-streaming sensors, it selects the minimum loaded sensor for its first sensor to transmit to. In this way, the loading of each sensor will be balanced. The above routing derives minimum balanced tree (MBT). This data structure will be adjusted locally while some sensors change their statuses in the network such that the control overhead needed to adjust is much less than to reconstruct all over again. Some results of simulated experimentations are shown in this paper.

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.