The Prediction Research on Multi Cameras Network Based on Rough Set Theory and Chaotic PSO

The multi cameras network design has the characteristics of the multiple parameters and constraint, the large amount of computation, which is an optimization strategy for complex problems in theory. We find usually the exact optimal solution to be require the very large computation. Therefore, the algorithm efficiency must be considered. In the paper, we employ the heuristic attribute reduction algorithms based on the discernibility matrix to reduce the attribute. Then predict the reduction attribute using the improved support vector machine. In order to obtain the optimal prediction accuracy, we use the chaotic particle swarm optimization (PSO) to prevent local optimization of SVM prediction models, which can make the prediction rate significantly increased. The analysis results show that the method can extract the basic rules of test data, compared with the single SVM method, the prediction error is reduced by more than 0.42%.