-
Next-Generation Applied Intelligence, 22nd International Conference on Industrial, Engineering and Other Applications of Applied Intelligent Systems, IEA/AIE 2009, Tainan, Taiwan, June 24-27, 2009. Proceedings; 01/2009
-
IJPRAI. 01/2006; 20:525-542.
-
18th International Conference on Pattern Recognition (ICPR 2006), 20-24 August 2006, Hong Kong, China; 01/2006
-
Int. J. Comput. Proc. Oriental Lang. 01/2001; 14:17-28.
-
[show abstract]
[hide abstract]
ABSTRACT: In this paper, we propose a GA-based unsupervised clustering technique that selects cluster centers directly from the data set, allowing it to speed up the fitness evaluation by constructing a look-up table in advance, saving the distances between all pairs of data points, and by using binary representation rather than string representation to encode a variable number of cluster centers. More effective versions of operators for reproduction, crossover, and mutation are introduced. Finally, the Davies-Bouldin index is employed to measure the validity of clusters. The development of our algorithm has demonstrated an ability to properly cluster a variety of data sets. The experimental results show that the proposed algorithm provides a more stable clustering performance in terms of number of clusters and clustering results. This results in considerable less computational time required, when compared to other GA-based clustering algorithms.
-
[show abstract]
[hide abstract]
ABSTRACT: This paper proposes a new clustering technique that is based on genetic algorithms, without the need for any GA operation. With the aid of an analysis of population Markov chains and some modifications to the genetic operations, the proposed technique markedly outperforms the existing conventional GA-based clustering methods. The proposed strategy adopts a modified version of Markov chain modeling introduced by Yong Gao et al. [16] to perform the evolutionary process. In the evolutionary process, offspring are simply produced according to the probabilities provided from Markov chain modeling, without any conventional genetic operators. Hence, a great deal of the processing time required by genetic operators can be eliminated. In the clustering procedure, the center candidates of the clusters are taken from the data set, so that we can in advance create a look up table that saves the distances between every pair of data points and prevent the repeated computation of distances in evaluating the fitness function. A binary representation is used to encode a certain set of cluster centers and the validity of the clusters is measured using the Davies-Bouldin index [12]. The experimental results indicate the superiority of the proposed algorithm over conventional genetic algorithms, and show that the proposed algorithms achieve better performance with less computational time than the conventional genetic algorithms.
-
[show abstract]
[hide abstract]
ABSTRACT: This paper presents an efficient and robust shape-based image retrieval system. The Mountain Climbing Sequence (MCS) [17] is used to represent the shape feature. This representation is invariant to translation, rotation, and scaling, and highly tolerate to occlusion and nosing. By taking into account the number of intersection points of each ray emanating from the centroid of the object with the object contour our method can effectively distinguish concavo-convex objects with different shapes. A branch-and-bound mechanism is applied to speed up the matching step. The results of our proposed method show a superior matching rate, even in the presence of a modest level of deformation or occlusion.
