This paper addresses an optimization problem using hybrid algorithms of the Sand Cat Swarm Optimization (SCSO) and Invasive Weed Optimization (IWO). In this study, the reproduction step in the IWO algorithm was incorporated after the initial population step of the SCSO. The proposed algorithm was compared against the following: Intersection Mutation Differential Evolution (IMDE), Differential Evolution (DE), SCSO, and Whale Optimization Algorithm (WOA), whereby the performance was tested on six benchmark functions using a 10-fold cross validation. The results indicate that the proposed algorithm yielded the optimal solution for two out of the six benchmark functions. Additionally, when compared with the other four chosen algorithms, it yielded the best overall results. The findings suggest that the proposed algorithm is able to generate solutions similar to those obtained from the previous methods, essentially for the continuous step function, the multimodal function, and the discontinuous step function.

This paper presents an improved K-means clustering algorithm that addresses the traditional algorithm’s sensitivity to outlier and susceptibility to local optima by introducing an adaptive weight adjustment mechanism. It employs an exponential decay function to dynamically reduce the feature weights of outlier data points, effectively suppressing outliers while preserving the structure of the normal data. The proposed method retains the computational efficiency of standard K-means. Key contributions include: (a) A novel distance-based weighting strategy that progressively reduces the influence of noisy points, mitigating the impact of outliers on clustering performance. (b) An innovative form of "local dimensionality reduction" for outlier points via weight decay, which interferes only with the feature space of noisy regions while preserving the global topological structure of clean data. Extensive experiments on three benchmark datasets Iris (4-dimensional, balanced classes), Wine (13-dimensional, correlated features), and Wisconsin Breast Cancer Diagnosis (30-dimensional, imbalanced data) demonstrate the effectiveness of the approach. Compared to standard K-means, the proposed algorithm achieves accuracy improvements of 7.47% on Iris, 13.89% on Wine, and 19% on WBCD. This adaptive strategy offers a practical and efficient solution for clustering in noisy, high-dimensional environments, without the added complexity of mixture models.

The proposed system represents an enhanced version in search of food of Sand Cat based on Levy distribution and Firework algorithm for the image classification of grape leaf diseases. In the preprocessing step, the proposed system utilizes convolution kernels to transform images into input data within the range of (0,1). Successively, the Levy distribution and Firework algorithm are incorporated into the SCSO model as an exploration search mechanism. The study employed a grape leaf dataset sourced from the Plant Village project (www.plantvillage.org), comprising 4062 labeled images measuring 256 by 256 pixels and categorized into 4 distinct classes: healthy, Black Rot, Black Measles, and Isariopsis leaf spot, which was utilized to evaluate the efficacy of the proposed system. The experimental findings demonstrate that the proposed system outperforms the analyses of VGG16, GLCM with SVM, Low contrast haze reduction-neighborhood component analysis with SVM, and SCSO.

In the realm of Data Mining, the pursuit of innovative classification methodologies remains crucial for advancing robust techniques in handling complex and diverse datasets. This paper explores the application of Differential Evolution (DE), a powerful optimization algorithm, as a unique and effective optimization-based standalone classifier. Our exploration focused on harnessing DE's intrinsic capabilities, adapting it into a classifier while preserving its distinctiveness. The fundamental principle of DE for classification involves iteratively optimizing the center point for each class using DE operators, which are specialized mechanisms for exploring and refining solutions, and utilizing these optimized points for making predictions. The effectiveness of our model was evaluated on ten classification datasets from the UCI Machine Learning Repository and compared against three other classification methods: KNN, ZMP, and BPNN. Experimental results underscore the competitive performance of our proposed model, emphasizing the potential of DE in effectively addressing classification challenges.

The proposed system represents an enhanced movement in search of food of Whale Optimization Algorithm (WOA), based on Levy distribution for image classification of grape leaf disease. In the preprocessing, the proposed system uses convolution kernels to transform images into input data within the range of (0,1). Thereafter, the Levy distribution is incorporated into the WOA model as an exploration search mechanism. A grape leaf dataset from the Plant Village project (www.plantvillage.org), consisting of 4062 labeled images with dimensions of 256 by 256 pixels and divided into four classes -healthy, Black Rot, Black Measles, and Isariopsis leaf spot -is used to evaluate the performance of the proposed system. Experimental results show that the proposed system is better than Visual Geometry Group (VGG16), Gray Level Co-occurrence Matrix (GLCM) with SVM, Low contrast haze reduction-neighborhood component analysis with SVM, and whale optimization algorithm (WOA).

Differential Evolution (DE) algorithms are widely recognized as effective metaheuristic techniques used to solve continuous optimization problems. However, DE algorithms encounter limitations in exploration and exploitation due to their sensitivity to parameter settings and reliance on mutation strategies. To overcome these drawbacks, we introduced a self-adaptive mutation and population improvement strategy in differential evolution, SAMPIDE. The algorithm adjusts the mutation operator automatically according to the successful mutation operator value from the previous generation. Additionally, to enhance the algorithm’s convergence, we adopted a random learning mechanism to update the population. SAMPIDEwas compared with nine state-of-the-art DE variants and four other metaheuristic algorithms across thirty-two benchmark functions. SAMPIDE consistently outperformed others on unimodal, step and noisy functions, while maintaining competitive performance on multimodal functions. Additionally, when tested on five real-world problems, SAMPIDE outperformed other metaheuristic algorithms. Thus, the self-adaptive mutation and population improvement strategy incorporated into SAMPIDE positioned it as a promising solution for a wide range of optimization problems.

The efficiency of a Differential Evolution (DE) algorithm largely depends on the control parameters of the mutation strategy. However, fixed-value control parameters are not effective for all types of optimization problems. Furthermore, DE search capability is often restricted, leading to limited exploration and poor exploitation when relying on a single strategy. These limitations cause DE algorithms to potentially miss promising regions, converge slowly, and stagnate in local optima. To address these drawbacks, we proposed a new Adaptive Differential Evolution Algorithm with Multiple Crossover Strategy Scheme (ADEMCS). We introduced an adaptive mutation strategy that enabled DE to adapt to specific optimization problems. Additionally, we augmented DE with a powerful local search ability: a hunting coordination operator from the reptile search algorithm for faster convergence. To validate ADEMCS effectiveness, we ran extensive experiments using 32 benchmark functions from CEC2015 and CEC2016. Our new algorithm outperformed nine state-of-the-art DE variants in terms of solution quality. The integration of the adaptive mutation strategy and the hunting coordination operator significantly enhanced DE's global and local search capabilities. Overall, ADEMCS represented a promising approach for optimization, offering adaptability and improved performance over existing variants. Doi: 10.28991/HIJ-2024-05-02-02 Full Text: PDF

We propose a new classification model—a new classification model for clustering overlapping problems based on K-Means and neural networks. K-means clustering algorithm belongs to unsupervised learning. It is a classic algorithm for solving clustering problems. Since this algorithm calculates its categories based on distance, the results tend to converge to the local optimal solution and have poor boundary clustering properties. The K-Means classification algorithm defines clusters by the distance between the cluster center value and the target object, and the optimal result is obtained through continuous iteration. Therefore, clustering results are overlapped, and there are often outliers that do not belong to the current cluster, resulting in unsatisfactory clustering results. Our model offers a new method to segment non-ideal data in overlapping regions. Since clustering algorithms cannot effectively identify and classify this part of the data, we split this part of the data and train it using a neural network. The results are then integrated into the clustered data. In the experiment, the k-fold cross-validation method ensures the model stability of the results. We used the accuracy to evaluate the quality of the model, and we used standard deviation and mean deviation to detect clustering results. Five sets of experimental data from the cross-experiment show that compared with the K-Means classification model, the accuracy of our model is effectively improved. Doi: 10.28991/HIJ-2024-05-03-012 Full Text: PDF

This research aims to improve the classification accuracy by modifying an original backpropagation neural network. In the proposed BPNN-ZMP, the learning rates were automatic tuned to improve the classification accuracy. Breast Cancer Coimbra dataset and Banknote Authentication dataset were used for testing the model performances. The results demonstrate that BPNN-ZMP improved over the original backpropagation neural network by 12.12 and 11.46% for Breast Cancer Coimbra dataset and Banknote Authentication dataset respectively. Although BPNN-ZMP could improve the model accuracy, the high accuracy in neural network backpropagation has been challenged in future work.

Travelling salesman problem is a special case of the vehicle routing problem. The objective of the travelling salesman problem is to find the shortest path for visiting every city without repeating city. Among metaheuristic algorithms, Ant System has been the most popular algorithm for solving the travelling salesman problem. However, Ant System has a disadvantage of often falling into local optimal solutions. This research proposed a modified Ant System with a modified pheromone density updating to reduce the rate of convergence. A threshold is also used to create a greater variety of routes. Moreover, the proposed algorithm used a Sweep Algorithm to generate initial population so that the next city to visit is close to each other. To prevent the ants from getting trapped at a local optimum, three types of local search, swap, insert, and reverse, are used to veer away the paths towards the higher pheromone density at a local optimum. The tested results of the proposed algorithm were compared to those of other three algorithms: GA-PSO-ACO, Hybrid VNS, and HAACO. On 13 out of 15 small- and medium-sized datasets, the proposed method outperformed or performed as well as the others.

The differential evolution algorithm has gained popularity for solving complex optimization problems because of its simplicity and efficiency. However, it has several drawbacks, such as a slow convergence rate, high sensitivity to the values of control parameters, and the ease of getting trapped in local optima. In order to overcome these drawbacks, this paper integrates three novel strategies into the original differential evolution. First, a population improvement strategy based on a multi-level sampling mechanism is used to accelerate convergence and increase the diversity of the population. Second, a new self-adaptive mutation strategy balances the exploration and exploitation abilities of the algorithm by dynamically determining an appropriate value of the mutation parameters; this improves the search ability and helps the algorithm escape from local optima when it gets stuck. Third, a new selection strategy guides the search to avoid local optima. Twelve benchmark functions of different characteristics are used to validate the performance of the proposed algorithm. The experimental results show that the proposed algorithm performs significantly better than the original DE in terms of the ability to locate the global optimum, convergence speed, and scalability. In addition, the proposed algorithm is able to find the global optimal solutions on 8 out of 12 benchmark functions, while 7 other well-established metaheuristic algorithms, namely NBOLDE, ODE, DE, SaDE, JADE, PSO, and GA, can obtain only 6, 2, 1, 1, 1, 1, and 1 functions, respectively. Doi: 10.28991/HIJ-2023-04-02-014 Full Text: PDF

This paper proposes an improvement of whale optimization algorithm for optimization problems. In this study, the Rao algorithm was improved by means of procedures of spiral updating position. The algorithm was tested on six benchmark problems and compared with differential evolution (DE), intersection mutation differential evolution (IMDE) algorithm, and whale optimization algorithm (WOA). The computation results illustrated that the proposed algorithm can produce optimal solutions for three out of six functions. Comparing to the other three algorithms, the proposed algorithm has provided the best results. The findings prove that the algorithm should be improved in this direction and show that the algorithm produces several solutions obtained by the previously published methods, especially for the continuous step function, the multimodal function, and the discontinuous step function.

This study presents a novel algorithm called hybrid-GA, which combines genetic algorithm (GA) with the Harris Hawks Optimization (HHO) algorithm to address the challenge of enhancing GA performance in solving optimization problems. While GA is known for its strong exploration capabilities, it often faces challenges in exploitation, limiting its ability to find global optimal solutions. The hybrid-GA algorithm aims to surpass existing methods by achieving a better balance between exploration and exploitation, resulting in improved solution quality, faster convergence, and enhanced exploration-exploitation ability. The algorithm effectiveness is demonstrated through experiments on six benchmark functions from CEC2017, where the hybrid-GA outperforms compared algorithms on five of six functions, showcasing its potential for enhancing GA performance in optimization problem-solving. These findings contribute to advancing the field by providing a promising solution to address the exploration-exploitation challenge in GA-based optimization.

This research introduces a novel approach, Discrete Differential Evolution (DDE), to address the Traveling Salesman Problem (TSP), a classic combinatorial optimization challenge with wide-ranging real-world applications. While initially designed for continuous optimization problems, the Differential Evolution algorithm has been adapted to suit the discrete nature of the TSP. Our approach involves several key modifications to the standard Differential Evolution algorithm. These include the creation of mutant vectors through random subtour optimization, the utilization of order crossover to generate trial vectors, and the application of a pool tournament population selection method to identify the most promising candidates. To assess the effectiveness of our proposed algorithm, we conducted comprehensive computational experiments across multiple TSP instances. Comparative analyses against other algorithms, such as the Hybrid Differential Evolution (HDE) and Discrete Particle Swarm Optimization (DPSO), consistently demonstrated that our proposed DDE outperformed the competitors on all Traveling Salesman test instances. This underscores the algorithm's superior performance, as evidenced by its consistently superior results in both the best and the mean solution quality, along with lower Relative Error (RE) values, signifying its proficiency in identifying routes closer to the optimum.

This paper proposed a modification of genetic algorithm with a new technique called the flexible crossover operation for solving the capacitated electric vehicle routing problem (CEVRP). The framework of this paper is based on the concept of the classic genetic algorithm (GA). To improve the performance of the genetic algorithm, two aspects have been proposed, 1) The Pre-Post mutation is present to obtain the ability of local search capacities. 2) The flexible crossover operation is proposed to enhance the diverse capabilities. The proposed algorithm has been evaluated and compared to the state-of-the-art algorithms. The experimental result on seven data sets demonstrated that the proposed algorithm is effective for solving a small and medium data set of capacitated electric vehicle routing problem.

Imbalanced data cause prediction inaccuracy of the classification model. Two types of techniques have been devised to address this problem: pre-processing data before training a classification model and adjusting the classification algorithm. This study, which introduced the adaptive learning rate into a backpropagation neural network algorithm, is of the latter type. The learning rate was adjusted in each iterative learning cycle: the learning rate is increased for the data class with fewer samples and decreased for the data class with more samples. K-fold cross-validation was used to test the effectiveness of the prediction model on 10 datasets. The results showed that the proposed ZMP algorithm outperformed the original backpropagation neural network on 6 datasets; the improvement ranged from 2.24% to 20.22%. Moreover, on the other 4 datasets, even though the proposed technique provided less accurate predictions, the differences were very slight.

Vehicle routing problem is a widely researched combinatorial optimization problem. We developed a hybrid of three strategies-a modified ant system, a sweep algorithm, and a path relinking-for solving a capacitated vehicle routing optimization problem, a vehicle routing problem with a capacity constraint. A sweep algorithm was used to generate initial solutions and assign customers to vehicles, followed by a modified ant system to generate new generations of good solutions. Path relinking was used for building a better solution (candidate) from a pair of guiding and initial solutions. Finally, a local search method-swap, insert, reverse and greedy search operations-was used to prevent solutions from getting trapped in a local minimum. Performance of the proposed algorithm was evaluated on three datasets from CVRPLIB. Our proposed method was at least competitive to state-of-the-art algorithms in terms of the total route lengths. It even surpassed the best-known solution in the P-n55-k8 instance. Our findings can lower the transportation cost by reducing the travelling distance and efficiently utilizing the vehicle capacity.

This research proposes a novel self-adaptive differential evolution algorithm for solving continuous optimization problems. This paper focuses on redesiging the self-adaptive strategy for the mutation parameters. The new mutation parameters adjust themselves to the current situation of the algorithm. When the search is stagnant, the first mutation parameter that scales the difference between the best vector and the target vector will be increased. In contrast, the second mutation parameter that scales the difference between two random target vectors will be decreased. On the other hand, when the search progresses well towards the global optimum, the algorithm will enhance the search of the surrounding space by doing the opposite of the above actions. The performance of the proposed self-adaptive differential evolution algorithm was evaluated and compared with the classic differential evolution algorithm on 7 benchmark functions. The experimental results showed that the proposed algorithm converged much faster than the classic differential evolution algorithm on all benchmark functions.

Clustering by the k-means is the most widely used method because of its ease of use. But the disadvantage of the k-means algorithm is that it relies on a random initialization. Therefore, the results obtained from each clustering are not stable depending on the starting point, affecting the results obtained in other applications. This paper, therefore, presents a method for determining the initialization of the k-means algorithm using the Data Distribution Guide (DDG). And use it as an aid in determining the starting point without random. Make the results of clustering always equal. And from the experimental results, We found that the accuracy obtained from clustering using the initialization from this method was good. Compared to the commonly used initialization designation.

This research modified a backpropagation learning algorithm in order to increase its ability to deal with imbalanced data problems. We used the backpropagation algorithm and a concept of multiple adaptive learning rates to train the feedforward neural network. Using multiple adaptive learning rates allowed us to achieve a classification model that had fewer problems when dealing with an imbalanced dataset. The experimental results showed that the proposed method performed significantly better than the conventional backpropagation neural network in all tests.