|
International Journal of Information Technology
Vol. 12 No. 3 2006 (Special Issue)
Guest Editorial
Special Issue on Intelligent Computing
Welcome to the special issue of IJIT on Intelligent Computing!
Intelligent computing is the study of adaptive mechanisms to enable or facilitate intelligent behavior in complex and changing environments. It has become one of the most promising techniques in the field of data processing as well as industrial applications. It is characterized by combining information science with biology. The concept of intelligent computing is quite fluid and perhaps this is the strength. It is seen to include a range of techniques such as artificial neural networks, evolutionary computing, swarm intelligence and fuzzy systems. Particularly in recent years, bio-inspired computing (e.g. Genetic Algorithm, Ant Colony Optimization, Life Systems, and Immune Systems, etc.) emerges as a key role in pursuing for novel technology. The resulting techniques vitalize life science engineering and daily life applications. Also, the methodologies of the fuzzy systems the rough set are also widely used in data processing and data management. Hence, the improvements of intelligent computing techniques and their applications in real-life world have become a hot issue. The theme of this special issue is to unify the picture of contemporary intelligent computing techniques as an integral concept that highlights the trends in advanced computational intelligence and bridges theoretical research with applications.
Fifteen papers are selected for this Special Issue after reviewed carefully by reviewers consisting of guest editors and external reviewers. Because of the diversity of the selected papers, it is difficult to go through in this editorial how the synthesis is achieved in each contribution and only a brief description is provided below to summarize the work of the papers.
1. Heuristic Algorithms
Seven papers that tackle various issues related to heuristic algorithms are selected in this special issue. While acknowledging its limited coverage, this collection of papers offers some interesting contributions.
- Weiqing Xiong, Liuvi Wang, and ChenYang Yan propose an ant colony system of binary network and combine this method with the Genetic Algorithm. This novel algorithm, which is called the binary ant colony evolutionary algorithm, is applied to some benchmark test functions and the multi 0/1 Knapsack problem. The results reveal that this algorithm shows better convergence speed and stability.
- Weijin Jiang, Yuhui Xu, and Yusheng Xu propose a novel application of neural network optimized design based on immune modulated symbiotic evolution on the basis of combining intergrowth evolving and immune adjustment. The results of simulation experiment applied in AGC-ASC system of the two stands reversing tandem cold mill show that this method is suited to the complicated climate, and has good convergence and resists disturbance.
- Qi Kang, Lei Wang, and Qidi Wu propose a novel fuzzy adaptive optimization strategy for the particle swarm algorithm. Firstly, the information of multi-optimum distribution state is introduced into the particle swarm movement programming. Secondly, a kind of fuzzy adaptive programming strategy based on double-variable and single-dimensional fuzzy control structure is proposed. The effectiveness of this approach is proven by simulation results of benchmark function optimization problems.
- Hao Mei, Yantao Tian, and Linan Zu propose a hybrid ant colony optimization algorithm for path planning of robot in dynamic environment that combines the characteristics of ant colony optimization (ACO) and artificial potential field (APF). ACO is used as global path planning and APF is used as local path planning. Moreover, some modifications are made to accommodate ACO to path planning problems. This algorithm is shown to satisfy the real-time demand by simulation results.
- Chaoxue Wang, Duwu Cui, Yikun Zhang, and Zhurong Wang propose a novel ant colony system which employs a candidate set strategy based on Delaunay triangulation (CSDT) and a self-adaptive mutation operator (SMO) for TSP (DSMACS). CSDT can limit the selection scope of ants at each step and thus substantially reduce the size of search space. SMO is designed to improve the global search ability of DSMACS by combining inversion and inserting mutation operator in genetic algorithm.
- Mohd Saufee Muhammad, Zuwairie Ibrahim, Osamu Ono, and Marzuki Khalid discuss the implementation ideas and experimental procedures to solve an engineering related combinatorial problem using DNA computing approach. An elevator scheduling problem is also chosen as a benchmark problem to be solved using this DNA computing techniques.
- Dong Hwa Kim applies PSO (Particle Swarm Optimization) and Euclidian data distance to the mutation procedure on GA’s differentiation to obtain global and local optimal solution together. This algorithm is proven by four benchmark test functions.
2. Information Processing & Clustering
Also, 8 papers concerning various issues of information processing and clustering techniques are selected:
- Lei Jia, LicaiYang, Qingjie Kong, and Shu Lin propose an artificial immune data clustering algorithm based on clone selective principle and immune network theory from the vertebrate immune systems. This algorithm does well in reducing the redundant information in clustered data, especially in dealing with the mass data clustering applications, where traditional clustering methods may be inefficient. This method is also applied to recognize traffic patterns in Urban Traffic Control problems.
- Feng Xia, Xiaohua Dai, Youxian Sun, and Jianxia Shou suggest a flexible control task model as an interface of control performance and scheduling decisions. Based on this model, a novel scheduling algorithm intended for control tasks is presented. A job skipping scheme is also introduced to handle the overload condition effectively. Preliminary simulations exhibit the benefits of the approach proposed in this paper, with comparison against traditional control systems design methodologies.
- Dechang Pi, Xiaolin Qin, and Qiang Wang propose an algorithm which is based on a dynamic tree for association rules clustering and applies it to association rules managing. Experiment shows that this algorithm can efficiently cluster the association rules for a user to understand.
- Yuntao Qian, Xiaoxu Du, and Qi Wang propose a semi-supervised hierarchical clustering algorithm for high dimensional data, which is based on the combination of semi-supervised clustering and dimensionality reduction. In order to achieve high harmony between dimensionality reduction and inherent cluster structure detection, the number of dimensions is reduced sequentially as the clusters are gradually formed in the hierarchical clustering procedure.
- Xiaobing Liu and Nan Zhang describe a novel behavior-based anti-Spam technology based on incremental immune-inspired clustering algorithm. An “internal image” network is used to represent the input data set in order to reduce data redundancy. At the same time, relevant information from the data set is extracted. Experimental evaluation shows that this novel approach provides significantly faster data summarization than completely re-clustering and the technology is reliable, efficient and scalable.
- Jose Aguilar, Juan Vizcarrondo, and Niriaska Perozo propose a verification method for the MASINA Methodology, which allows the specification of Multi-Agent Systems (MAS). The MAS verification design described in this paper is divided into two levels: the Macro Level verifies the design of each subsystem that makes up the MAS, and the Micro Level validates the design from the point of view of models.
- Jifeng Chen, Li Zhu, Junyi Shen, Zhihai Wang, and Xinjun Wang analyze the properties and disadvantages of predicate constraint solving technique and present a novel test data generating approach, using predicate constraint solving techniques for program execution.
- Duo Chen, Duwu Cui, Chaoxue Wang, and Zhurong Wang applie the rough set theory to clustering analysis. The clustering data set is mapped as the decision table through introducing a decision attribute. Also, the categorical similarity measure based on Euclidean distance is suggested. Theoretical analysis and experimental results indicate that this algorithm is valid.
Finally, we like to would like to thank all our contributing authors for their dedicated work.
Jun Zhang, Guest Editor
Department of Computer Science
Sun Yat-sen University
Guangzhou P.R. China
junzhang@ieee.orgJun Zhang
|