همکاری مصنوعی کلنی زنبور برای بهینه سازی خوشه ها با استفاده از SPNN

This paper deals with methods for clustering of continuous signals such as time series data sets. Centers of classes are determined with the help of the neural network with process input, which is an extension of the traditional artificial neural network into the time domain. Collaborative Artificial Bee Colony algorithm is based on the search of food behaviour of honey bees for training in a non-trajectory way. An Enhancement has been done to the original Artificial Bee Colony (ABC) algorithm and was used to discover suitable domain specific architectures. The C-ABC has great explorative search features and better convergence compared to the original algorithm and it was proved empirically that it avoids local minima by promoting exploration of the search space. In SPNN(Self Organizing Process Neural Network), the inputs and weights are related to instantaneous conditions. The proposed algorithm results in clustering the data sets with reduced error rate and better convergence rate. The tests are conducted on empirical data in matlab.

In neural networks research at present, the most popular and effective model is a feedforward neural network. It is quite successful in many domains, such as pattern recognition, classification and clustering, adaptive control and learning, etc. Clustering is a common problem in signal processing and combinatorial analysis. When the class number is unknown, the method of merging samples into classes is called “clustering”. For clustering, the classification structure of research objects does not need to be known beforehand, and it can be classified according to similarities among the research objects, which are not restricted by the current level of study of research objects and prior knowledge. The feedforward neural network (such as a self-organizing mapping neural network withoutteaching) adopting the self-organizing competitive learning algorithm with process input is a young approach to obtain the perfect cluster, which is broadly applied to many fields including data mining, association analysis, etc.

This work proposed a novel learning scheme for process neural networks based Gaussian mixture weight functions and Collaborative Artificial Bee Colony Algorithm (C-ABC). The weight function was expressed as the generalized Gaussian mixture functions. The structure and parameters were optimized by C-ABC. According to the results of experiments, C-ABC had best performance on time-series prediction and pattern recognition than BPNN, ABC-LM and PSO-LM. In spite of having complex calculations the best performance is attained. This paper has presented a new clustering method based on Collaborative Artificial Bee Colony Algorithm (C-ABC). The method employs the Algorithm to search for the set of cluster centers that minimizes a given clustering metric. One of the advantages of the proposed method is that it does not become trapped at locally optimal solutions. This is due to the ability of the C-ABC Algorithm to perform local and global search simultaneously experimental results for different data sets have demonstrated that the proposed method produces better performances than those clustering algorithm.