Multi-Layer Perceptron Neural Network Utilizing Adaptive Best-Mass Gravitational Search Algorithm to Classify Sonar Dataset

In this paper, a new Multi-Layer Perceptron Neural Network (MLP NN) classifier is proposed for classifying sonar targets and non-targets from the acoustic backscattered signals. Besides capabilities of MLP NNs, it uses Back Propagation (BP) Gradient Descent (GD) training; therefore, NNs face with not only impertinent classification accuracy but also getting stuck in local minima as well low-convergence speed. To lift defections, study Adaptive Best Mass Gravitational Search Algorithm (ABGSA) to train NN. This algorithm develops marginal disadvantage GSA using best-collected masses within iterations expediting exploitation phase. test classifier, along GSA, GD, GA, PSO compound method (PSOGSA) via three datasets various dimensions will be assessed. Assessed metrics include convergence speed, fail probability minimum accuracy. Finally, practical application assumed network classifies dataset. dataset consists echoes six different objects: four two non-targets. Results indicate that proposes better output terms aforementioned criteria than whole benchmarks.