New approach to solve Uncapacitated Facility Location Problem using Continuous Hopfield Network and modified K-means clustering

Our work addresses a new method to handle the Uncapacited Facility Location Problem using neural networks and K-means clustering algorithm. In the first phase, we use the Maximum Stable Set Problem (MSSP) and the Continuous Hopfield Network (CHN) to determine the initial number of facilities to establish and their optimal locations. The second one consists to, based on results of the first step and using an improved K-means clustering technique, assign customers to the corresponding facilities in order to achieve the lowest possible overall cost of the process. Experiments, evaluations and comparison with the exact solution of the Uncapacited Facility Location Problem are carried out with various data sets, and it yields good results regarding quality of solution and the execution time.