Don-Lin Yang

An efficient Fuzzy C-Means clustering algorithm

The Fuzzy C-Means (FCM) algorithm is commonly used for clustering. The performance of the FCM algorithm depends on the selection of the initial cluster center and/or the initial membership value. If a good initial cluster center that is close to the actual final cluster center can be found, the FCM algorithm will converge very quickly and the processing time can be drastically reduced. The authors propose a novel algorithm for efficient clustering. This algorithm is a modified FCM called the psFCM algorithm, which significantly reduces the computation time required to partition a dataset into desired clusters. We find the actual cluster center by using a simplified set of the original complete dataset. It refines the initial value of the FCM algorithm to speed up the convergence time. Our experiments show that the proposed psFCM algorithm is on average four times faster than the original FCM algorithm. We also demonstrate that the quality of the proposed psFCM algorithm is the same as the FCM algorithm.

Applying UML and software simulation for process definition, verification, and validation

Process definition, verification, and validation are recognized as critical elements in software process improvement, whereas CMMI is a process improvement approach that provides organizations with the essential elements of effective processes. Organizations must define their own processes to meet the requirements of CMMI. A friendly, unambiguous process modeling language and tool are thus very important for organizations to define, verify, and validate the processes. Nevertheless, hardly has any research yet been done on how to embed CMMI process area goals into process definition stage to satisfy organization process improvement. In this research, we propose a UML-based approach to define, verify, and validate an organizations process. Our approach can also be applied to a process learning environment for students and project members.

A generalized processor mapping technique for array redistribution

In many scientific applications, array redistribution is usually required to enhance data locality and reduce remote memory access in many parallel programs on distributed memory multicomputers. Since the redistribution is performed at runtime, there is a performance trade-off between the efficiency of the new data decomposition for a subsequent phase of an algorithm and the cost of redistributing data among processors. In this paper, we present a generalized processor mapping technique to minimize the amount of data exchange for BLOCK-CYCLIC(kr) to BLOCK-CYCLIC(r) array redistribution and vice versa. The main idea of the generalized processor mapping technique is first to develop mapping functions for computing a new rank of each destination processor. Based on the mapping functions, a new logical sequence of destination processors can be derived. The new logical processor sequence is then used to minimize the amount of data exchange in a redistribution. The generalized processor mapping technique can handle array redistribution with arbitrary source and destination processor sets and can be applied to multidimensional array redistribution. We present a theoretical model to analyze the performance improvement of the generalized processor mapping technique. To evaluate the performance of the proposed technique, we have implemented the generalized processor mapping technique on an IBM SP2 parallel machine. The experimental results show that the generalized processor mapping technique can provide performance improvement over a wide range of redistribution problems.

Efficient approaches for materialized views selection in a data warehouse

View materialization is an effective method to increase query efficiency in a data warehouse and improve OLAP query performance. However, one encounters the problem of space insufficiency if all possible views are materialized in advance. Reducing query time by means of selecting a proper set of materialized views with a lower cost is crucial for efficient data warehousing. In addition, the costs of data warehouse creation, query, and maintenance have to be taken into account while views are materialized. In this paper, we propose efficient algorithms to select a proper set of materialized views, constrained by storage and cost considerations, to help speed up the entire data warehousing process. We derive a cost model for data warehouse query and maintenance as well as efficient view selection algorithms that effectively exploit the gain and loss metrics. The main contribution of our paper is to speed up the selection process of materialized views. Concurrently, this will greatly reduce the overall cost of data warehouse query and maintenance.

Using Data Mining to Study Upstream and Downstream Causal Relationship in Stock Market

To understand the causal relationship of stock market is always a top priority for investors. Most investors use some fundamental knowledge and basic analysis techniques to analyze or predict the trends. However, there are always some other factors beyond our control or unexpected events that might affect the stock market one way or the other. After working on data mining with good results, we found inter-transaction mining can help answer the above questions in a systemic way. Our experiments show that causal relationship between upstream and downstream stocks do exist. To simplify our discussion, we focus on the electrical industrial stocks.

Image Processing and Image Mining using Decision Trees

Valuable information can be hidden in images, however, few research discuss data mining on them. In this paper, we propose a general framework based on the decision tree for mining and processing image data. Pixel-wised image features were extracted and transformed into a database-like table which allows various data mining algorithms to make explorations on it. Each tuple of the transformed table has a feature descriptor formed by a set of features in conjunction with the target label of a particular pixel. With the label feature, we can adopt the decision tree induction to realize relationships between attributes and the target label from image pixels, and to construct a model for pixel-wised image processing according to a given training image dataset. Both experimental and theoretical analyses were performed in this study. Their results show that the proposed model can be very efficient and effective for image processing and image mining. It is anticipated that by using the proposed model, various existing data mining and image processing methods could be worked on together in different ways. Our model can also be used to create new image processing methodologies, refine existing image processing methods, or act as a powerful image filter.

Efficient Mining of Generalized Negative Association Rules

Most association rule mining research focuses on finding positive relationships between items. However, many studies in intelligent data analysis indicate that negative association rules are as important as positive ones. Therefore, we propose a method improved upon the traditional negative association rule mining. Our method mainly decreases the huge computing cost of mining negative association rules and reduces most non-interesting negative rules. By using a taxonomy tree that was obtained previously, we can diminish computing costs, through negative interestingness measures, we can quickly extract negative association data from the database.

An efficient hash-based method for discovering the maximal frequent set

The association rule mining can be divided into two steps. The first step is to find out all frequent itemsets, whose occurrences are greater than or equal to the user-specified threshold. The second step is to generate reliable association rules based on all frequent itemsets found in the first step. Identifying all frequent itemsets in a large database dominates the overall performance in the association rule mining. In this paper, we propose an efficient hash-based method, HMFS, for discovering the maximal frequent itemsets. The HMFS method combines the advantages of both the DHP (Direct Hashing and Pruning) and the Pincer-Search algorithms. The combination leads to two advantages. First, the HMFS method, in general, can reduce the number of database scans. Second, the HMFS can filter the infrequent candidate itemsets and can use the filtered itemsets to find the maximal frequent itemsets. These two advantages can reduce the overall computing time of finding the maximal frequent itemsets. In addition, the HMFS method also provides an efficient mechanism to construct the maximal frequent candidate itemsets to reduce the search space. We have implemented the HMFS method along with the DHP and the Pincer-Search algorithms on a Pentium III 800 MHz PC. The experimental results show that the HMFS method has better performance than the DHP and the Pincer-Search algorithms for most of test cases. In particular, our method has significant improvement over the DHP and the Pincer-Search algorithms when the size of a database is large and the length of the longest itemset is relatively long.

Pervasive health service system: insights on the development of a grid-based personal health service system

Although medical technologies developed in the twenty-first century have successfully increased mans life span, the pressure of modern life has consequently brought many modern civilization diseases and chronic illness. When all these problems are tackled by hospitals, they will consume considerable amount of medical resources. Alternatively, providing health care services at home is an important issue for improving personal health and save hospital resources. In this paper, we present an ongoing project that designs and implements a pervasive health service infrastructure based on the grid system which is integrated with the P2Ps resource sharing mechanism, to provide the personal health service. The personal health status is recorded, monitored, and even mined in/from the proposed pervasive health service system for preventive medicine. Additionally, wireless sensor equipments for mobile personal health services are also integrated into the pervasive health service system, in order to construct a situation-aware, context-aware and environment-aware mobile-health-service platform.

Implementation of Wiki-Based Knowledge Management Systems for Small Research Groups

Due to the soaring of knowledge-based economy and the blooming growth of e-business, the way to effectively accumulate and manage valuable data of knowledge has become an important issue for all the industry. There are many knowledge management systems built for large corporate companies. However, not many open source systems are available for small research communities. Currently, Wiki has not only changed the conception of building knowledge in an on-going process, but also become a popular tool of sharing knowledge worldwide. In this paper, we adopt Wiki tools, Tiddly-Wiki and Media-Wiki, to build two useful platforms as knowledge management systems for small research groups less than twenty members. Furthermore, we also provide integration service via the Update Server, which supports knowledge update, coherence editing, and knowledge sharing in order to satisfy the requirements of personal and group research. Our prototype intends to help students quickly grasp the expertise of their specific domain knowledge as new comers and become experienced researchers after graduation. We try to make this project as a continuous effort to accumulate more useful knowledge in a collaborative manner as required by the researchers.