Sam Hsu

The Impact of Gene Selection on Imbalanced Microarray Expression Data

Microarray experiments usually output small volumes but high dimensional data. Selecting a number of genes relevant to the tasks at hand is usually one of the most important steps for the expression data analysis. While numerous researches have demonstrated the effectiveness of gene selection from different perspectives, existing endeavors, unfortunately, ignore the data imbalance reality, where one type of samples (e.g., cancer tissues) may be significantly fewer than the other (e.g., normal tissues). In this paper, we carry out a systematic study to investigate the impact of gene selection on imbalanced microarray data. Our objective is to understand that if gene selection is applied to imbalanced expression data, what kind of consequences it may bring to the final results? For this purpose, we apply five gene selection measures to eleven microarray datasets, and employ four learning methods to build classification models from the data containing selected genes only. Our study will bring important findings and draw numerous conclusions on (1) the impact of gene selection on imbalanced data, and (2) behaviors of different learning methods on the selected data.

Experimental analysis of application-level intrusion detection algorithms

Intrusion Detection System (IDS) plays a very important role on information security. In this paper, we present an application-level intrusion detection algorithm named Graph-based Sequence-Learning Algorithm (GSLA). GSLA includes data pre-processing, normal profile construction and session marking. In GSLA, the normal profile is built through a session-learning method, which is used to determine an anomaly session. We conduct experiments and evaluate the performance of GSLA with other conventional algorithms, such as Markov Chain Model (MM) and K-means Algorithm. The results show that GSLA improves the effectiveness of anomaly detection.

Design and development of a hybrid instruction model for a new teaching paradigm

In recent years, many universities have sought to develop distance learning courses and programs that are delivered through the Web. The promising results achieved by many of these projects lead to a question: how can we integrate the best features of Web-based learning into a conventional classroom-based model of instruction? To answer this question, the Department of Computer Science and Engineering and the Department of Languages and Linguistics at Florida Atlantic University, USA, are jointly working on a pilot project that integrates conventional classroom teaching and Web-based distance learning technologies to form a hybrid instruction model for a teaching paradigm that can be easily applied toward learner-centered education. This paper explains the motivation for this project and describes its main technical aspects. A number of the pedagogical and practical issues related to the proposed model are reviewed and the results obtained in its first trial are discussed. Conclusions drawn from a student exit survey and from classroom experience are presented at the end of the article with recommendations for future implementations of a hybrid instructional model.

Design and development of a Web-based academic advising system

Academic advising is an important and time-consuming task and different tools and techniques can be used to make it an effective and efficient process. This paper describes the design and development of a Web-based advising system that supplements the conventional advising process. The systems goals include: to minimize repetitive tasks performed by advisors, to encourage students to adopt a proactive attitude towards advising, to make advising-related information available to remote students in a single place, in electronic format, and to minimize inconsistencies in the advising process. The system supports three different types of users (students, advisors, and secretaries), each of which has different privileges and allowed operations. Student users may use the system to find relevant advising-related information, such as course descriptions and advising FAQs. They can also ask the system which course(s) to take next, based on the classes they have already taken.

FEATURE SELECTION FOR DATASETS WITH IMBALANCED CLASS DISTRIBUTIONS

Feature selection for supervised learning concerns the problem of selecting a number of important features (w.r.t. the class labels) for the purposes of training accurate prediction models. Traditional feature selection methods, however, fail to take the sample distributions into consideration which may lead to poor prediction for minority class examples. Due to the sophistication and the cost involved in the data collection process, many applications, such as biomedical research, commonly face biased data collections with one class of examples (e.g., diseased samples) significantly less than other classes (e.g., normal samples). For these applications, the minority class examples, such as disease samples, credit card frauds, and network intrusions, are only a small portion of the data but deserve full attention for accurate prediction. In this paper, we propose three filtering techniques, Higher Weight (HW), Differential Minority Repeat (DMR) and Balanced Minority Repeat (BMR), to identify important features from datasets with biased sample distribution. Experimental comparisons with the ReliefF method on five datasets demonstrate the effectiveness of the proposed methods in selecting informative features for accurate prediction of minority class examples.

Towards an Internet‐based Education Model for Caribbean Countries

Abstract Advances in the development of the Internet infrastructure can be used within developing countries to enhance the delivery of high quality education to their citizens. The mechanisms used to deliver educational material, over the Internet, range from the very sophisticated Virtual Classrooms (VCs) in which students in cyberspace interact in near real‐time with instructors at remote sites, to the basic non real‐time delivery of lecture contents. In Caribbean territories where the concentration of high‐quality teachers and teaching facilities are in and around urban cities, VCs can be used to enhance the access of students in remote areas to the materials, experiences, and facilities provided at the urban centers. We present a model for virtual classrooms, which is specifically geared toward Caribbean territories. This model leverages the usage of alternative low‐cost technology to offer VC environment to teaching institutions in remote rural districts via the Internet.

Buffer allocation optimization in ATM switching networks using ALOPEX algorithm

Abstract We describe the use of a stochastic algorithm, called ALOPEX, which could be implemented in VLSI for optimizing the buffer allocation process in ATM switching networks. We present the results of computer simulations for buffer allocation in ATM switching networks using the ALOPEX algorithm. The algorithm uses a scalar cost function which is a measure of global performance. The ALOPEX works by broadcasting the global cost function to all neural processors in the neural network. Since each neural processor solely depends on the global cost function no interaction is needed between the neural processors and the algorithm is more amenable to massively parallel implementation. The application of the ALOPEX algorithm for the buffer allocation optimization in ATM networks assumes limited buffer capacity. The proposed ALOPEX-based approach takes advantage of the favorable control characteristics of the algorithm such as high adaptability and high speed collective computing power for effective buffer utilization. The proposed model uses complete sharing buffer allocation strategy and enhances its performance for high traffic loads by regulating the buffer allocation process dynamically.

Towards interoperability of UML tools for exchanging high-fidelity diagrams

In todays global software engineering projects, where development activities are distributed geographically and temporally, it is increasingly important for CASE tools to maintain the information (both syntactic and semantic) captured in the design models. The Unified Modeling Language (UML) is the de facto standard for modeling software applications and UML diagrams serve as graphical documentations of the software system. The interoperability of UML modeling tools is important in supporting the model exchange. Tool interoperability is often implemented using XML Metadata Interchange (XMI). Unfortunately, there is a loss of fidelity of the design documentation when transforming between UML and XMI due to the compatibility of different versions of UML, XMI and add-on proprietary information, which hinder reuse. This paper reports on an ongoing study evaluating the interoperability of UML modeling tools by assessing the quality of XMI documents representing the design. Case studies in the paper demonstrate a framework of preserving the fidelity of UML models data when importing and exporting different UML models in a distributed heterogeneous environment.

HWSAM: a Web-based automated homework submission system

A Web-based automated homework submission system has been developed to allow students to submit their homework across the Internet. This system is composed of three modules: a front-end GUI interface module, an online homework submission acceptance module, and a backend GUI interface module. The front-end GUI interface module includes a standard Web browser and a set of CGI programs and HTML documents to interact with the users for prompting them to submit their files. The on-line homework submission acceptance module accepts files received from students and organizes them into hierarchical directory structures based on courses, assignments, and student IDs. The back-end GUI interface is structured similarly to the front-end interface; however, it is to be used by instructors for grading purposes. All homework files submitted will be confirmed in two ways. First, it is confirmed at the time of submission by displaying a message on the screen. Second, it e-mails the filename to the submitting student for a personal record. Each file received is timestamped automatically at the time of submission by the host operating system. Files of any data type, text or binary can be submitted in the same manner. Ease of use is a major design goal for this system. For students, menus and browse windows are used to navigate the Web pages for submitting homework and checking grades received. The only typing needed is for entering in student names and IDs. For instructors, grading can be done on-line via the HTML forms, or the student files can be downloaded and then graded later.

Feature selection with biased sample distributions

Feature selection concerns the problem of selecting a number of important features (w.r.t. the class labels) in order to build accurate prediction models. Traditional feature selection methods, however, fail to take the sample distributions into the consideration which may lead to poor predictions for minority class examples. Due to the sophistication and the cost involved in the data collection process, many applications, such as Biomedical research, commonly face biased data collections with one class of examples (e.g., diseased samples) significantly less than other classes (e.g., normal samples). For these applications, the minority class examples, such as disease samples, credit card frauds, and network intrusions, are only a small portion of the data collections but deserve full attentions for accurate prediction. In this paper, we propose three filtering techniques, Higher Weight (HW), Differential Minority Repeat (DMR) and Balanced Minority Repeat (BMR), to identify important features from biased data collections. Experimental comparisons with the ReliefF method on five datasets demonstrate the effectiveness of the proposed methods in selecting informative features from data with biased sample distributions.