An effective gbest-guided gravitational search algorithm for real-parameter optimization and its application in training of feedforward neural networks

Crocodile optimization algorithm for solving real-world optimization problems

Nature-inspired bionic algorithms have become one of the most fascinating techniques in computational intelligence research, and have shown great potential in real-world challenging problems for their simplicity and flexibility. This paper proposes a novel nature-inspired algorithm, called the crocodile optimization algorithm (COA), which mimics the hunting strategies of crocodiles. In COA, the hunting behavior of crocodiles includes premeditation and waiting hunting. The premeditation behavior is an important hunting way for crocodiles to hide themselves from their prey and to explore more potential areas, and the waiting hunting behavior is another means by which crocodiles make surprise attacks on their prey that appears in their hunting range. The performance of the proposed COA is validated by comparing it with other competitor algorithms on 29 standard test functions and 5 real-world engineering optimization problems. The experimental results show that the comprehensive performance of COA outperforms both of its similar variants and most of other state-of-the-art algorithms, in terms of solution accuracy, robustness and convergence speed. Statistical tests also validate the potential applications of the proposed algorithm.

Effective PID controller design using a novel hybrid algorithm for high order systems

This paper discusses the merging of two optimization algorithms, atom search optimization and particle swarm optimization, to create a hybrid algorithm called hybrid atom search particle swarm optimization (h-ASPSO). Atom search optimization is an algorithm inspired by the movement of atoms in nature, which employs interaction forces and neighbor interaction to guide each atom in the population. On the other hand, particle swarm optimization is a swarm intelligence algorithm that uses a population of particles to search for the optimal solution through a social learning process. The proposed algorithm aims to reach exploration-exploitation balance to improve search efficiency. The efficacy of h-ASPSO has been demonstrated in improving the time-domain performance of two high-order real-world engineering problems: the design of a proportional-integral-derivative controller for an automatic voltage regulator and a doubly fed induction generator-based wind turbine systems. The results show that h-ASPSO outperformed the original atom search optimization in terms of convergence speed and quality of solution and can provide more promising results for different high-order engineering systems without significantly increasing the computational cost. The promise of the proposed method is further demonstrated using other available competitive methods that are utilized for the automatic voltage regulator and a doubly fed induction generator-based wind turbine systems.

Application of Meta-Heuristic Algorithms for Training Neural Networks and Deep Learning Architectures: A Comprehensive Review

The learning process and hyper-parameter optimization of artificial neural networks (ANNs) and deep learning (DL) architectures is considered one of the most challenging machine learning problems. Several past studies have used gradient-based back propagation methods to train DL architectures. However, gradient-based methods have major drawbacks such as stucking at local minimums in multi-objective cost functions, expensive execution time due to calculating gradient information with thousands of iterations and needing the cost functions to be continuous. Since training the ANNs and DLs is an NP-hard optimization problem, their structure and parameters optimization using the meta-heuristic (MH) algorithms has been considerably raised. MH algorithms can accurately formulate the optimal estimation of DL components (such as hyper-parameter, weights, number of layers, number of neurons, learning rate, etc.). This paper provides a comprehensive review of the optimization of ANNs and DLs using MH algorithms. In this paper, we have reviewed the latest developments in the use of MH algorithms in the DL and ANN methods, presented their disadvantages and advantages, and pointed out some research directions to fill the gaps between MHs and DL methods. Moreover, it has been explained that the evolutionary hybrid architecture still has limited applicability in the literature. Also, this paper classifies the latest MH algorithms in the literature to demonstrate their effectiveness in DL and ANN training for various applications. Most researchers tend to extend novel hybrid algorithms by combining MHs to optimize the hyper-parameters of DLs and ANNs. The development of hybrid MHs helps improving algorithms performance and capable of solving complex optimization problems. In general, the optimal performance of the MHs should be able to achieve a suitable trade-off between exploration and exploitation features. Hence, this paper tries to summarize various MH algorithms in terms of the convergence trend, exploration, exploitation, and the ability to avoid local minima. The integration of MH with DLs is expected to accelerate the training process in the coming few years. However, relevant publications in this way are still rare.

Training a Feedforward Neural Network Using Hybrid Gravitational Search Algorithm with Dynamic Multiswarm Particle Swarm Optimization

One of the most well-known methods for solving real-world and complex optimization problems is the gravitational search algorithm (GSA). The gravitational search technique suffers from a sluggish convergence rate and weak local search capabilities while solving complicated optimization problems. A unique hybrid population-based strategy is designed to tackle the problem by combining dynamic multiswarm particle swarm optimization with gravitational search algorithm (GSADMSPSO). In this manuscript, GSADMSPSO is used as novel training techniques for Feedforward Neural Networks (FNNs) in order to test the algorithm's efficiency in decreasing the issues of local minima trapping and existing evolutionary learning methods' poor convergence rate. A novel method GSADMSPSO distributes the primary population of masses into smaller subswarms, according to the proposed algorithm, and also stabilizes them by offering a new neighborhood plan. At this time, each agent (particle) increases its position and velocity by using the suggested algorithm's global search capability. The fundamental concept is to combine GSA's ability with DMSPSO's to improve the performance of a given algorithm's exploration and exploitation. The suggested algorithm's performance on a range of well-known benchmark test functions, GSA, and its variations is compared. The results of the experiments suggest that the proposed method outperforms the other variants in terms of convergence speed and avoiding local minima; FNNs are being trained.

Path planning for the autonomous robots using modified grey wolf optimization approach

Path planning for robots plays a vital role to seek the most feasible path due to power requirement, environmental factors and other limitations. The path planning for the autonomous robots is tedious task as the robot needs to locate a suitable path to move between the source and destination points with multifaceted nature. In this paper, we introduced a new technique named modified grey wolf optimization (MGWO) algorithm to solve the path planning problem for multi-robots. MGWO is modified version of conventional grey wolf optimization (GWO) that belongs to the category of metaheuristic algorithms. This has gained wide popularity for an optimization of different parameters in the discrete search space to solve various problems. The prime goal of the proposed methodology is to determine the optimal path while maintaining a sufficient distance from other objects and moving robots. In MGWO method, omega wolves are treated equally as those of delta wolves in exploration process that helps in escalating the convergence speed and minimizing the execution time. The simulation results show that MGWO gives satisfactory performance than other state of art methods for path planning of multiple mobile robots. The performance of the proposed method is compared with the standard evolutionary algorithms viz., Particle Swarm Optimization (PSO), Intelligent BAT Algorithm (IBA), Grey Wolf Optimization (GWO), and Variable Weight Grey Wolf Optimization (VW-GWO) and yielded better results than all of these.

Removal of computed tomography ring artifacts via radial basis function artificial neural networks

Ring artifacts in computed tomography (CT) images are caused by the undesirable response of detector pixels, which leads to the degradation of CT images. Accordingly, it affects the image interpretation, post-processing, and quantitative analysis. In this study, a radial basis function neural network (RBFNN) was used to remove ring artifacts. The proposed method employs polar coordinate transformation. First, ring artifacts were transformed into linear artifacts by polar coordinate transformation. Then, smoothing operators were applied to locate these artifacts exactly. Subsequently, RBFNN was operated on each linear artifact. The neuron numbers of the input, hidden, and output layers of the neural network were 8, 40, and 1, respectively. Neurons in the input layer were selected according to the characteristics of the artifact itself and its relationship with the surrounding normal pixels. For the training of the neural network, a hybrid of adaptive gradient descent algorithm (AGDA) and gravitational search algorithm (GSA) was adopted. After the corrected image was obtained using the updated neural network, the inverse coordinate transformation was implemented. The experimental data were divided into simulated ring artifacts and real ring artifacts, which were based on brain and abdomen CT images. Compared with current artifact removal methods, the proposed method removed ring artifacts more effectively and retained the maximum detail of normal tissues. In addition, for index analysis, the performance of proposed method was superior to that of the other methods.

Combining a gravitational search algorithm, particle swarm optimization, and fuzzy rules to improve the classification performance of a feed-forward neural network

BACKGROUND AND OBJECTIVE

A feed-forward neural network (FNN) is a type of artificial neural network that has been widely used in medical diagnosis, data mining, stock market analysis, and other fields. Many studies have used FNN to develop medical decision-making systems to assist doctors in clinical diagnosis. The aim of the learning process in FNN is to find the best combination of connection weights and biases to achieve the minimum error. However, in many cases, FNNs converge to the local optimum but not the global optimum. Using open disease datasets, the purpose of this study was to optimize the connection weights and biases of the FNN to minimize the error and improve the accuracy of disease diagnosis.

METHOD

In this study, the chronic kidney disease (CKD) and mesothelioma (MES) disease datasets from the University of California Irvine (UCI) machine learning repository were used as research objects. This study applied the FNN to learn the features of each datum and used particle swarm optimization (PSO) and a gravitational search algorithm (GSA) to optimize the weights and biases of the FNN classifiers based on the algorithms inspired by the observation of natural phenomena. Moreover, fuzzy rules were used to optimize the parameters of the GSA to improve the performance of the algorithm in the classifier.

RESULTS

When applied to the CKD dataset, the accuracies of PSO and GSA were 99%. By using fuzzy rules to optimize the GSA parameter, the accuracy of fuzzy-GSA was 99.25%. The accuracies of the combined algorithms PSO-GSA and fuzzy-PSO-GSA reached 100%. In the MES disease dataset, all methods exhibited good performance with 100% accuracy.

CONCLUSIONS

This study used PSO, GSA, fuzzy-GSA, PSO-GSA, and fuzzy-PSO-GSA on CKD and MES disease datasets to identify the disease, and the performance of different algorithms was explored. Compared with other methods in the literature, our proposed method achieved higher accuracy.

A QSAR model for predicting antidiabetic activity of dipeptidyl peptidase-IV inhibitors by enhanced binary gravitational search algorithm

Time-varying binary gravitational search algorithm (TVBGSA) is proposed for predicting antidiabetic activity of 134 dipeptidyl peptidase-IV (DPP-IV) inhibitors. To improve the performance of the binary gravitational search algorithm (BGSA) method, we propose a dynamic time-varying transfer function. A new control parameter, μ , is added in the original transfer function as a time-varying variable. The TVBGSA-based model was internally and externally validated based on Qint2 , QLGO2 , QBoot2 , MSEtrain , Qext2 , MSEtest , Y-randomization test, and applicability domain evaluation. The validation results indicate that the proposed TVBGSA model is robust and not due to chance correlation. The descriptor selection and prediction performance of TVBGSA outperform BGSA method. TVBGSA shows higher Qint2 of 0.957, QLGO2 of 0.951, QBoot2 of 0.954, Qext2 of 0.938, and lower MSEtrain and MSEtest compared to obtained results by BGSA, indicating the best prediction performance of the proposed TVBGSA model. The results clearly reveal that the proposed TVBGSA method is useful for constructing reliable and robust QSARs for predicting antidiabetic activity of DPP-IV inhibitors prior to designing and experimental synthesizing of new DPP-IV inhibitors.