Numerical optimization and feed-forward neural networks training using an improved optimization algorithm: multiple leader salp swarm algorithm

An Optimized Continuous Dragonfly Algorithm Using Hill Climbing Local Search to Tackle the Low Exploitation Problem

“…From Tables 7, 8, 9, and 10 it can be deduced that the proposed optimized DA algorithm has a higher effectiveness as compared to multiple other swarm intelligence algorithms in training artificial neural networks. For the iris, glass, and breast cancer datasets, the ANN trained by the proposed optimized DA achieves a higher accuracy than all other swarm intelligence algorithms used to train ANNs in [47], [48], and [49]. For the balloon dataset, the accuracy obtained by the ANN trained by the optimized DA is 100% which is the same as the ANNs trained by several other algorithms.…”

“…6, it can be seen that the optimized DA algorithm converges to the optimal solution at around iteration 15 while the original DA converges at around iteration 18. Moreover, it can be seen that the optimized DA In order to have a fair comparison, the optimized DA algorithm is used for training ANNs with the same architecture as the other works in [47], [48], and [49] for the four datasets. For the iris dataset, the following architecture is used: one input layer with four neurons, one hidden layer with nine neurons, and one output layer with three neurons, for the balloon dataset, the following architecture: one input layer with four neurons, one hidden layer with nine neurons, and one output layer with one neuron, for the glass dataset, the architecture used is: one input layer with nine neurons, one hidden layer with 19 neurons, and one output layer with one neuron, and for the breast cancer dataset, the following architecture is used: one input layer with nine neurons, one hidden layer with 19 neurons, and one output layer with one neuron.…”

“…To further analyse the performance of the Optimized DA algorithm, its performance in training ANNs using the iris, balloon, glass, and breast cancer datasets is compared to that of other swarm intelligence algorithms. The results obtained when several other swarm intelligence algorithms are used for training ANNs using the iris, balloon, glass, and breast cancer datasets are taken from [47], [48], and [49]. Specifically, the results obtained when ANNs are trained using Ant Colony Optimization (ACO), Ant Lion Optimization (ALO), Bat Algorithm (BAT), Biogeography-based Optimization (BBO), Cuckoo Search (CS), Differential Evolution (DE), Elephant Herding Optimization (EHO), Evolution Strategy (ES), Genetic Algorithm (GA), Gravitational Search Algorithm (GSA), Grey Wolf Optimization (GWO), Harmony Search (HS), Moth-Flame Optimization (MFO), Multiple Sequence Alignment (MSA), Particle Swarm Optimization (PSO), Sine Cosine Algorithm (SCA), Whale Optimization Algorithm (WOA), and Hybrid ABC-DA (HAD) algorithms are taken from [47], those trained by Hybrid Nelder-Mead and DA (INMDA) and Population-Based Incremental Learning (PBIL) algorithms are taken from [48], and those trained by Multiple Leader Salp Swarm Algorithm (MLSSA) and Salp Swarm Algorithm (SSA) algorithms are taken from [49].…”

“…In order to have a fair comparison, the optimized DA is used to train ANNs with the same architecture as in [47], [48], and [49]. Hence, the number of neurons in the hidden layer is changed to nine for the iris dataset and balloon dataset, and to 19 for the glass and breast cancer datasets.…”

Manta ray foraging optimization algorithm–based feedforward neural network for electric energy consumption forecasting

“…The results showed the superiority of the proposed approach compared to the results of other algorithms. 53 Researchers have also focused on solving the problem of training ANNs using multiple leader salp swarm algorithm, 54 hybrid particle swarm optimization-genetic algorithm (PSO-GA), 55 Dragonfly Algorithm, 56 social-learning particle swarm optimization algorithm (SL-PSO), 57 tree-seed algorithm (TSA), 58 a novel Hybrid Sine Cosine Algorithm, 59 Artificial Algae Algorithm, 60 and PSO. 61 Studies in the literature indicate that different optimization algorithms have been used in the training of ANNs.…”