Chuleui Hong

An innovative method for data and software integration in SaaS

Recently the main trend in providing software services has been shifting from an ASP (application service provider)-oriented to a SaaS (software as a service). ASP is a software service model in which the service is provided on a one by one basis according its ownership, while SaaS is a software service model in which the service is provided virtually on a one by one basis, but physically all at once. In a SaaS environment, all users can access the system via Internet without any software installation-examples include Google and Amazon. Now, more companies are shifting their business software service from ASP to SaaS. However effecting the transition of the existing software and data from ASP to SaaS is not an easy task. First, we have to solve the problem of the integration of data for different forms of software, because each data set consists of different data types. Second, the software integration must support a user customizing interface for various users on the Web. Almost all users want customized services, but those require high costs. In this paper we propose a novel method for transferring the existing business software to integrated software that can be used in the SaaS environment. We use a master table and master code to implement the integrated system. The master table is based on the project master table, and other user information tables are connected to collect the necessary information. All information about the project is stored in each column of the project master table. The master table can integrate various software databases. By using this novel methodology, the existing ASP-based software and data can be effectively transferred to the SaaS environment.

Distributed channel routing using genetic algorithm

Known as an NP-Complete problem, the channel routing problem is very important in the automatic layout design of VLSI circuit and printed circuit boards. In this paper, a distributed genetic algorithm for this channel routing problem is presented in MPI environments. Each node executes genetic operations to own sub-population and communicates synchronously with other nodes to form the global population. The experimental results show that the proposed algorithm maintains the convergence properties of sequential genetic algorithm.

Distributed simulated annealing for composite stock cutting problem

The composite stock cutting problem is defined as allocating rectangular and irregular patterns onto a large composite stock sheet of finite dimensions in such a way that the resulting scrap will be minimized. In this paper, we propose distributed simulated annealing for this composite stock cutting problem. The new cost error tolerant scheme relaxes synchronization and does move generations asynchronously in a dynamically changed stream length to keep the convergence property of the sequential annealing. This paper also proposes the efficient data structures for pattern related information.

A novel approach to function approximation: Adaptive multimodule regression networks

In a function approximation task using neural network, there exist many cases in which the distributions of data are so complex that the regression with single network does not perform the given task well. Employing multiple modules that performs regression in different regions respectively may be a reasonable solution for this case. This paper proposes a new adaptive modular architecture for regression model, and simulates its performance on difficult problems that are not easily approximated using traditional neural network learning algorithms. Regression modules are added as learning proceeds, depending on the local distribution of data. The use of a local distribution that captures the underlying local structure of the data offers the advantage of adaptive regression.

Asynchronous distributed genetic algorithm for optimal channel routing

This paper presents a distributed genetic algorithm for the channel routing problem in MPI environments. This system is implemented on a network of personal computers running Linux operating system connected via 10Mbps Ethernet. Each slave processor generates its own sub-population using genetic operations and communicates with the master processor in an asynchronous manner to form the global population. The experimental results show that the proposed algorithm maintains the convergence properties of sequential genetic algorithm while it achieves linear speedup as the nets of the channel routing and the number of computing processors increase.

The Artificial Neural Network based Electric Power Demand Forecast using a Season and Weather Informations

This paper proposes the new electric power demand forecast model which is based on an artificial neural network and considers time and weather factors. Time factors are selected by measuring the autocorrelation coefficients of load demand in summer and winter seasons. Weather factors are selected by using Pearson correlation coefficient The important weather factors are temperature and dew point because the correlation coefficients between these factors and load demand are much higher than those of the other factors such as humidities, air pressures and wind speeds. The experimental results show that the proposed model using time and seasonal weather factors improves the load demand forecasts to a great extent.

A Multimedia Authoring and Virtual Collaboration System Supporting Multi-Conferences for E-Learning

We are proposing the multimedia presentation authoring and virtual collaboration system that produces multimedia contents and enables multiple users who are dispersed in locally and timely to collaborate by the Internet. The proposed system consists of two parts—multimedia presentation authoring and recordable virtual collaboration tools. The authoring tool makes it possible to create and edit multimedia presentations that integrate diverse media types including images, video, sound, and texts for e-learning. Media objects are synchronized with the temporal and spatial information using SMIL defined by W3C. The collaboration tool categorizes users by the interested conference and any users can create new topics or join the existing topics by validating user’s access right freely in a conference they belong. Users can participate in more than one topic simultaneously, so they can inquire and get a valuable knowledge on-line in one topic and participate in the other topic more informed and intelligent. The produced multimedia presentation may be provided to the users through the conference before the discussion begins. Users can use text along with associated symbols such as arrows and polygons over the presented images during the discussion. Users’ opinions along with symbols are recorded to XML database.

A Web-Based Virtual Collaboration System in Multiple Conferences

We are proposing the web-based virtual collaboration system where multiple users are collaborated dispersed in locally and timely by the Internet. The proposed system categorizes users by the interested conference and each user is restricted to access the system resources by the access control. Any users can crate new topics or join the existing topics by validating user’s access right freely in a conference they belong. Users can participate in more than one topic simultaneously, so they can inquire and get a valuable knowledge on-line in one topic and participate in the other topic more informed and intelligently. They can also save time by attending multiple topics at the same time. Conferences are performed by using power point slides and associated spatial data elements from the participants. Associated spatial data elements can be symbols such as arrows or polygons along with text. The content of each conference – spatial data elements from the participants with respect to certain topic – can be recorded into an XML file and archived for future reference.

A Multi-classifier System Using Mean Field Genetic Algorithm

This paper presents an approach for building a multi-classifier system in Mean Field Genetic Algorithm (MGA) based inductive learning environments. Several base classifiers are combined with a meta-classifier that learns the bias of base classifiers so that it can draw a decision by combining predictions made by base classifiers. MGA is a hybrid algorithm of Mean Field Annealing (MFA) and Simulated annealing-like Genetic Algorithm (SGA). The proposed MGA combines the benefit of rapid convergence property of MFA and the effective genetic operations of SGA.

Learning the bias of a classifier in a GA-Based inductive learning environment

We have explored a meta-learning approach to improve the prediction accuracy of a classification system. In the meta-learning approach, a meta-classifier that learns the bias of a classifier is obtained so that it can evaluate the prediction made by the classifier for a given example and thereby improve the overall performance of a classification system. The paper discusses our meta-learning approach in details and presents some empirical results that show the improvement we can achieve with the meta-learning approach in a GA-based inductive learning environment.