Ant Colony Optimization Algorithm in the Hopfield Neural Network for Agricultural Soil Fertility Reverse Analysis

Abstract

The Boolean satisfiability problem is one of the most important decision problems in mathematical logic and computational science for determining whether or not a solution to a Boolean formula exists. The Hopfield neural network (HNN) is a major type of artificial neural network (ANN), and it is widely used to solve various optimization and decision problems due to its energy minimization mechanism. Existing models that incorporate a standalone network-projected non-versatile framework as a fundamental HNN employ random search in their training stages and are sometimes trapped at the local optimal solution. In this study, the ant colony optimization (ACO) algorithm as a novel variant of the probabilistic metaheuristic algorithm inspired by the behavior of real ants is incorporated into the training phase of HNN to accelerate the training process for random Boolean k satisfiability reverse analysis based on logic mining. The performance of the proposed hybrid model is evaluated in terms of the robustness and accuracy of the induced logic obtained by using the Agricultural Soil Fertility Data Set. Experimental simulation results reveal that ACO can effectively work with HNN for Random 3 satisfiability reverse analysis with 87.5% classification accuracy.