To improve the performance of the predator-prey cellular genetic algorithm and distinguish different populations in genotype, a multi-species predator-prey cellular genetic algorithm with linear mapping is proposed. All individuals are divided into two parts, denoted predators and preys. The viability of individual is proportional to its fitness. A mapping matrix is applied to the process of calculating the fitness of population to change the mapping relationship between genotype and phenotype and make different species carry with different genetic information. During the evolution, species use different crossover methods and adjust the mapping matrix coefficients based on the dispersion degree of populations to control the evolution direction of the population and thus the ability of escaping from local optimum is enhanced. Compared with some other similar algorithms on several low and high dimension typical complicated functions, the proposed algorithm shows fine optimizing performance in global convergence.

In collaborative filtering, sparsity in ratings makes inaccurate neighborhood formation, thereby resulting in poor recommendations. To address this issue, a method of neighborhood formation, two-phase neighbor selection method (TPNS), is proposed. The definition of neighbor tendency is given. Based on the neighbor tendency, the preliminary neighborhood is formed. Then, the equivalence relation similarity is applied to modify the preliminary neighborhood, which makes the neighborhood formation more accurate. Experimental results on MovieLens dataset show that compared with the existingalgorithms, TPNS performs better in the application of personalized recommendation.

To weaken the influence of the insufficient datasets and noises on the clustering analysis, a clustering algorithm, transfer generalized fuzzy C-means with improved fuzzy partitions (T-GIFP-FCM) is proposed based on the FCM framework-based clustering algorithm GIFP-FCM. By leveraging the historical knowledge in the related scene (domain) , the performance of T-GIFP-FCM is enhanced. Even if the data in the current scene are not enough, the promising clustering results can be obtained. The experimental results show the proposed algorithm has better performance compared with the traditional algorithms in situations of insufficient data.

To improve the node energy utilization and extend the lifetime of the network, a clustering node sleep scheduling algorithm with particle swarm optimization(CNSS-PSO) in wireless sensor networks (WSNs) is presented by combining clustering algorithm and sleep scheduling algorithm. In CNSS-PSO, the binary encoding mechanism is used and mutation and crossover operators of the genetic algorithm are introduced. The maintenance of network coverage and the optimization goals of the reduction of energy consumption are taken into account, and a corresponding discrete particle swarm optimization method is constructed. Simulation results show that CNSS-PSO improves the effectiveness of the network energy consumption and it has better performance on the network coverage maintenance with effectively extending the lifetime of the network.