Self-organization Membrane Computing-based Attribute Weights Optimization for Case-based Reasoning Model

To solve the problem of the distribution of the attribute weights in the retrieval process based on case-based reasoning (CBR), a self-organizing membrane computing method was proposed to calculate the attribute weights. Firstly, the fitness function was established to evaluate the rationality of the weight distribution, and the one level membrane structure with cell type and the membrane rule with selection, crossover, mutation and two-way communication were designed to search the optimal weight object set iteratively. Then, according to the fitness function and rules designed for training. The number of basic membrane and the reasonable value of attribute weights was obtained. Finally, 5 regression data sets from UCI and the dissolved oxygen concentration data from the wastewater treatment process were used to carry out a comparison experiment. The results show that the proposed method can effectively reduce the fitting error of regression and receive the reasonable distribution of attribute weights, thus to further improve the solution to the performance of CBR model.