A Novel Method for Initializing Populations Using the Metropolis–Hastings (MH) Technique

This chapter presents a new population initialization method for metaheuristic algorithms. This approach generates the initial set of candidate solutions by sampling the objective function using the Metropolis–Hastings technique. Unlike many existing initialization methods that mainly focus on spatial distribution, this algorithm ensures that the initial points closely align with the salient values of the objective function to be optimized. This distinctive feature allows this approach to target promising regions within the search space, improving the probability of identifying the global optimum. Consequently, this method demonstrates accelerated convergence and improved solution quality compared to traditional techniques. This initialization method is perfectly integrated into the classical differential evolution algorithm to demonstrate its effectiveness. Performance evaluation involves testing the entire system on benchmark functions extracted from various data sets. The experimental results underline the superior convergence speed and higher solution quality offered by this technique, highlighting its effectiveness compared to similar approaches.