Understanding social and environmental behavior patterns through hybrid neural swarm clustering

Most of the clustering problems can be reformulated as combinational optimization problems. It is not easy to search for the global solution for combinational optimization problems. In this paper, we use the Discrete Hopfield Neural Network (DHN) to solve the Bipartition Clustering Problem (BCP) and combine Particle Swarm Optimization (PSO) to search for the global solution. Firstly, the BCP is reformulated into an integer optimization problem. Secondly, to ensure the local solution of BCP in convergence and stability, some rules of DHN are designed to solve the Integer Optimization Problem(IOP). Finally, PSO is proposed to reset the neuron states of DHN until the global solution of BCP is achieved. Numerical and real-world experiments are conducted to evaluate the validity and feasibility of the proposed method. The experimental results show that our method achieves better clustering results on different datasets and problem instances, and has higher accuracy and stability compared to traditional methods.