Search

Differential evolution (DE) has attracted widespread attention due to its outstanding optimization performance and ease of operation, but it cannot avoid the dilemmas of premature convergence or stagnation when faced with complex optimization problems. To reduce the probability of such difficulties for DE, we sort out the factors that influence the balance between global exploration and local exploitation in the DE algorithm, and we design a novel DE variant (abbreviated as PISRDE) by integrating the corresponding influence factors through a periodic intervention mechanism and a systematic regulation mechanism. The periodic intervention mechanism divides the optimization operations of PISRDE into routine operation and intervention operation, and it balances global exploration and local exploitation at the macro level by executing the two operations alternately. The systematic regulation mechanism treats the involved optimization strategies and parameter settings as an organic system for targeted design, to balance global exploration and local exploitation at the meso or micro level. To evaluate and verify the optimization performance of PISRDE, we employ seven DE variants with excellent optimization performance to conduct comparison experiments on the IEEE CEC 2014 and IEEE CEC 2017 benchmarks. The comparison results indicate that PISRDE outperforms all competitors overall, and its relative advantage is even more significant when dealing with high-dimensional and complex optimization problems. Schematic design philosophy of PISRDE [ABSTRACT FROM AUTHOR]

Differential evolution (DE) has been proven as a simple yet powerful meta-heuristic algorithm on tackling continuous optimization problems. Nevertheless most existing DE methods still suffer from certain drawbacks including the use of ineffective mechanisms to adjust mutation strategies and their control parameters that may possibly mislead the search directions, and also the lack of intelligent guidance and reset mechanisms to escape from local optima. Therefore, to enhance the adaptability of DE-based search frameworks and the robustness on optimizing complex problems full of local optima, an adaptive and guided differential evolution (AdaGuiDE) algorithm is proposed. Essentially, the adaptability of the AdaGuiDE search framework is enhanced by three schemes to iteratively refine the search behaviour at two different levels. At the macroscopic level, the AdaGuiDE search framework revises the existing adaptive mechanism for selecting appropriate DE search strategies by counting the actual contributions in terms of solution quality. In addition, the adaption strategy is extended to the microscopic level where a penalty-based guided DE search is employed to guide the search escaping from local optima through temporarily penalizing the local optima and their neighborhood. Furthermore, a systematic boundary revision scheme is introduced to dynamically adjust the search boundary for locating any potential regions of interest during the search. For a rigorous evaluation of the proposed search framework, the AdaGuiDE algorithm is compared against other well-known meta-heuristic approaches on three sets of benchmark functions involving different dimensions in which the AdaGuiDE algorithm attained remarkable results especially on the high-dimensional and complex optimization problems. More importantly, the proposed AdaGuiDE framework shed lights on many possible directions to further enhance the adaptability of the underlying DE-based search strategies in tackling many challenging real-world applications. [ABSTRACT FROM AUTHOR]

As the complexity and quantity of network data continue to increase, accurate and efficient anomaly detection methods become critical. Deep learning-based methods are suitable for real-time detection because they leverage neural networks to efficiently process massive amounts of data. However, for complex network environments and unknown threats, it is difficult to acquire balanced datasets for training, resulting in low model accuracy. Moreover, in a large-scale network environment, the model training process is complicated and resource-consuming, ignoring the important information hidden behind the data and poor scalability. To address these issues, we develop a novel network anomaly detection method that integrates fused feature imageization with an enhanced extreme learning machine, termed GMD-DELM. The network data stream features are transformed into images by an adaptive transformation method, generating a feature representation with highly enhanced data recognition capability. In addition, a ResNeXt network embedded with an attention mechanism is used to extract high-level features from images, enhancing the ability of deep learning networks to extract important features from network streams. Finally, we implement a network anomaly detection method established on an improved adaptive differential evolution kernel extreme learning machine. The experimental results demonstrate that the proposed model achieves notable enhancements achieved by the proposed model in reducing feature redundancy and improving accuracy compared to existing network anomaly detection models. Furthermore, our model exhibits improved stability and robustness in detecting corrupted network data containing noise. [ABSTRACT FROM AUTHOR]

This paper deals with an N P -hard problem called the colored balanced traveling salesman problem (CBTSP), which is a variation of colored traveling salesman problem (CTSP) which in turn is a variation of multiple traveling salesman problem. To effectively solve this problem, an approach based on evolution strategy is proposed where mutation operator is designed taking into consideration the characteristics of CBTSP. The results obtained through our approach have been compared with the results of the novel genetic algorithm (NGA) which is the best approach available in the literature. The results show that our approach surpasses the results reported in the literature for majority of the instances in shorter amount of time. [ABSTRACT FROM AUTHOR]

In the real world, data often exhibits high-dimensional and complex characteristics. In addition, an object may correspond to multiple class labels. Therefore, filtering and processing such data has become a hot research topic. Multi-label feature selection is one of the key preprocessing techniques for effectively solving redundant information. However, due to the high complexity of multi-label learning feature selection and involving optimization of multiple objectives, careful consideration is required to achieve effective optimization. To address these issues, this paper proposes a multiple reference points-based multi-objective feature selection for multi-label learning. Firstly, the neighborhood crossover strategy utilizes individuals in the same neighborhood for crossover, thereby enhancing the search effect within that decision space. The distribution-based mutation strategy adjusts the mutation probability of different features based on their distribution in selected candidate solutions to increase the likelihood of potentially useful features being selected. Secondly, the multiple reference points-based selection strategy uses a scalar function to uniformly evaluate offspring individuals and control their distance from each other through multiple reference points. This approach helps in selecting convergent solutions with balanced diversity. Finally, experimental results demonstrate that the proposed algorithm attains an optimal performance of 83% across fourteen widely used datasets, outperforming ten state-of-the-art algorithms in terms of effectiveness and performance in multi-label feature selection. [ABSTRACT FROM AUTHOR]

In the real world, data often exhibits high-dimensional and complex characteristics. In addition, an object may correspond to multiple class labels. Therefore, filtering and processing such data has become a hot research topic. Multi-label feature selection is one of the key preprocessing techniques for effectively solving redundant information. However, due to the high complexity of multi-label learning feature selection and involving optimization of multiple objectives, careful consideration is required to achieve effective optimization. To address these issues, this paper proposes a multiple reference points-based multi-objective feature selection for multi-label learning. Firstly, the neighborhood crossover strategy utilizes individuals in the same neighborhood for crossover, thereby enhancing the search effect within that decision space. The distribution-based mutation strategy adjusts the mutation probability of different features based on their distribution in selected candidate solutions to increase the likelihood of potentially useful features being selected. Secondly, the multiple reference points-based selection strategy uses a scalar function to uniformly evaluate offspring individuals and control their distance from each other through multiple reference points. This approach helps in selecting convergent solutions with balanced diversity. Finally, experimental results demonstrate that the proposed algorithm attains an optimal performance of 83% across fourteen widely used datasets, outperforming ten state-of-the-art algorithms in terms of effectiveness and performance in multi-label feature selection. [ABSTRACT FROM AUTHOR]

The opposition-based differential evolution (ODE) cannot adaptively adjust the number of individuals partake opposition-based learning, which makes it difficult to solve complex optimization problems. In this manuscript, we present an innovative approach for the treatment of variable population ODE (SASODE) by leveraging on adaptive parameters. The core idea of SASODE is to assign a jumping rate to each individual in the population, which is the key parameter that determines whether an individual enters a subpopulation or not. The initial rate assignment relies on the empirical mean of a normal distribution. During the iterative process, the mean is adjusted adaptively by taking into account the historical information of the individuals retained from the preceding generation. At the same time, the variation of this mean directly lead to changing the jumping rate of individuals and thus to adjusting the subpopulation size. In addition, the constant c and the Lehmer mean together maintain a balance between exploration and exploitation of SASODE. Experimental results show that the algorithm ranks first in the Wilcoxon test on 61 benchmarks and three optimization problems in three dimensions. Then, we confirm that SASODE can achieve an accuracy of 96% or even higher on the feature selection problem. Therefore, SASODE outperforms the other state-of-the-art algorithms compared in terms of convergence rate and accuracy. [ABSTRACT FROM AUTHOR]

Differential evolution (DE) is an evolutionary algorithm that is straightforward but effective and has been satisfactorily implemented in a variety of fields. However, the DE relies heavily on the mutation strategy, and utilizing different mutation strategies can boost DE's performance at different phases during the evolution. This paper proposes a DE variant with the strategy transformation (qSTDE) based on a quartile. The proposed qSTDE uses feedback information from individuals and the entire population distribution to determine the position ofmutation strategy transformation, balancing theDE's exploration and exploitation capabilities. Individuals are first ranked in accordance with their fitness values, and then the ranked population is separated into four sub-populations using quartile points. Next, the Euclidean distances between the best individual and four individuals that are randomly selected from the four sub-populations are calculated, as well as their Euclidean distances to the searching space ratios. Further, the evolution process is completely separated into two phases, and the transformation position is determined through the comparison of ratios for a given threshold. Finally, different mutation strategies are used in different stages depending on the distinction between the two stages and the transformation position. The proposed qSTDE is evaluated on CEC2005 and CEC2014 benchmark test functions. The experimental results show that the qSTDE is efficient and competitive contrasted to the state-of-the-art DE variants. [ABSTRACT FROM AUTHOR]

Economic dispatch (ED) is a key foundational issue for optimal power system operation and scheduling control. It is a complex multi-constraint optimization problem, especially taking into account the valve-point effects of thermal power generators. As the power system continues to grow in size, the ED problem becomes more sophisticated and the solution space will have more local extrema, which makes the solution methods more prone to premature convergence. Thus, the existing methods encounter difficulties in achieving satisfactory solutions. To address this issue, this study presents an exponential hybrid mutation differential evolution (EHMDE), which utilizes two improved strategies including exponential population size reduction and hybrid mutation operation to adaptively equilibrate exploitation and exploration during the iteration process. The former strategy can maintain population diversity to avoid getting stuck in a local optimum in the preceding period and enhance the convergence speed in the later period by reducing the population size progressively. The latter strategy can explore wide search ranges and aggregate the individuals by two mutation operators EHMDE/current-to-rand/1 and EHMDE/pbest/1 based on a variation probability. Simulation results of 23 benchmark functions and five ED cases verify the superiority of EHMDE over other peer methods. Furthermore, they also demonstrate that these two improved strategies work well together to strengthen EHMDE. [ABSTRACT FROM AUTHOR]

This study investigates the joint replenishment and delivery problem taking trade credits the delay in payment permitted by suppliers and capacity-constrained into consideration (CJRD-TC). The CJRD-TC aims to minimize the total system costs by finding a reasonable replenishment and delivery schedule policy. In this paper, an opposite-based RUNge Kutta optimization algorithm (OBRUN) is proposed to compare with other intelligent algorithms, including the differential evolution (DE), RUNge Kutta optimization algorithm (RUN), sine cosine algorithm (SCA), and adaptive differential evolution with optimal external archive (JADE). The proposed algorithm is tested on 14 optimization functions. The values of average, standard deviation, and the best are the lowest for 12 functions, which demonstrates OBRUN is superior to other algorithms. We also provide graphs of search history, 2D views of functions, trajectory curve, average fitness history, and convergence curve of OBRUN, the balance between exploration and exploitation over the course of iterations can be observed. The convergence curves of five algorithms are also depicted, which show OBRUN has a fast convergence rate and accuracy than other algorithms. We conduct numerical experiments, parameter tuning, and sensitivity analysis for each of the small-scale, medium-scale, and large-scale number of items. The OBRUN can always find better solutions with the lowest mean values and the best-found total cost results than other algorithms. Indicators of the average optimality gap and the optimality gap dominate by the value of zero. These computational results have demonstrated OBRUN is a suitable and effective candidate for managers to solve the practical CJRD-TC problems. [ABSTRACT FROM AUTHOR]

Bilevel evolutionary algorithms (BLEAs) are a plausible approach for bilevel optimization. However, these algorithms require many fitness evaluations (FEs) and might become unusable if the fitness evaluations are computationally expensive. Therefore, reducing the number of FEs is crucial for designing a BLEA for expensive bilevel optimization (EBLOP). The surrogate-assisted optimization and knowledge transfer mechanisms used in BLEAs have been proven to reduce the number of FEs. This paper proposes a surrogate-assisted bilevel improved multioperator differential evolution algorithm (SA-BL-IMODE), which integrates a classification model-based assisted preselection and environment selection strategy (CPES) for EBLOP. In CPES, promising candidate solutions are prescreened by a classification model, filtering out some unpromising candidate solutions without performing FEs, thus improving the algorithm's performance. Moreover, the classification model also assists environment selection by directly discarding the unpromising offspring solutions before FEs are performed, thus reducing the number of FEs in each iteration. Additionally, an enhanced direct neighbor solution transfer (EDST) mechanism is proposed to identify and utilize the interactions between upper-level and lower-level variables for acquiring knowledge, improving knowledge quality, and reducing the number of FEs further. Experimental studies on two test suite benchmark problems are conducted, and the proposed method is compared with nine state-of-the-art algorithms. The experimental results demonstrate the effectiveness of the proposed mechanisms and show that SA-BL-IMODE has a significant advantage over existing algorithms for expensive bilevel optimization. [ABSTRACT FROM AUTHOR]

Multi-task optimization is an emerging research topic in evolutionary computation, which aims to solve multiple optimization tasks simultaneously through knowledge transfer. However, existing multi-task evolutionary algorithms suffer from the re-evaluation problem, leading to unnecessary consumption of computing resources. To address this issue, a non-revisiting framework is proposed, which allows the non-revisiting scheme to be aided by historical information during the evolutionary search process. Moreover, an individual updating strategy is designed to improve the search efficiency of the algorithm and enhance the ability to escape local optima. Furthermore, a parallel scheme of the proposed framework is developedNational Frontiers Science Center for Industrial Intelligence and Systems Optimizationcomputation time on the CUDA architecture. To evaluate the effectiveness of the proposed framework, it is integrated with success-history based adaptive differential evolution. A comparative study of the proposed algorithm with eight state-of-the-art multi-task evolutionary algorithms is performed on nine benchmark problems. The experimental results demonstrate that the proposed algorithm outperforms the existing algorithms, highlighting its potential for solving multi-task optimization problems. [ABSTRACT FROM AUTHOR]

The width spread is one of the key indices affecting hot rolling processes and product quality. The traditional Shibahara spread prediction model (SSM) does not take into account the wear and interference of production equipment. As a result, the prediction accuracy cannot satisfy the high precision and high reliability requirements. To improve the prediction accuracy, first, this paper analyzes the strip deformation mechanism and the source of prediction errors and proposes an improved Shibahara spread model (ISSM) based on wear and interference. Second, a Bayesian-optimized differential evolution and adaptive gradient descent (BDE-AGD) algorithm is proposed for ISSM parameter optimization. This algorithm adopts the Bayesian algorithm to optimize the differential evolution (BDE) algorithm in stage 1, which improves the global search ability. Furthermore, the adaptive strategy is used for gradient descent (AGD) to improve the local search capability in stage 2. Finally, the experimental results show that the BDE-AGD algorithm reduces the error by 21.9% and 3.1% compared with DE-AGD and BDE-GD, respectively. This shows that BDE-AGD has a better global optimization capability. In addition, the prediction precision of ISSM is improved by 9.77% compared with the traditional SSM. Moreover, the precision of ISSM is improved by 8.56% on average compared with seven machine learning algorithms. This shows that ISSM optimized by BDE-AGD can achieve high-precision prediction of strip spread. [ABSTRACT FROM AUTHOR]

The unit commitment problem (UCP), which includes the unit schedule and power dispatch, is a nonlinear high-dimensional and highly constrained mixed-integer combinatorial optimization problem. One challenge herein is to obtain high-quality solutions considering various constraints. Developing a competitive hybrid method is a mainstream study goal in this field, which has focused on the unit schedule optimization but less on power dispatch. Inspired by the advantage of genetic algorithms (GAs) in solving combinational optimization problems and the characteristic of grey prediction evolution algorithm (GPE) with strong exploration ability, this paper proposes a novel hybrid GA and GPE method, termed hGAGPE, to solve the UCP. In hGAGPE, GPE, as a novel real parameter stochastic search algorithm based on the grey prediction theory for data mining, is first employed to solve the power dispatch of the UCP. Meanwhile, the unit schedule is performed by the popular GA. Additionally, some heuristic repair mechanisms based on the priority list and an elite selection mechanism are incorporated to enhance the performance of hGAGPE. The proposed hGAGPE is evaluated on six test systems with generating units in the range of 10 to 100 during a 24-h scheduling period. The numerical results demonstrate the feasibility and effectiveness of hGAGPE in comparison with other existing approaches. [ABSTRACT FROM AUTHOR]

Clustering for probability density functions (CDF) can be categorized as non-fuzzy and fuzzy approaches. Regarding the second approach, the iterative refinement technique has been used for searching the optimal partition. This method could be easily trapped at a local optimum. In order to find the global optimum, a meta-heuristic optimization (MO) algorithm must be incorporated into the fuzzy CDF problem. However, no research utilizing MO to solve the fuzzy CDF problem has been proposed so far due to the lack of a reasonable encoding for converting a fuzzy clustering solution to a chromosome. To address this shortcoming, a new definition called Gaussian prototype is defined first. This type of prototype is capable of accurately representing the cluster without being overly complex. As a result, prototypes' information can be easily integrated into the chromosome via a novel prototype-based encoding method. Second, a new objective function is introduced to evaluate a fuzzy CDF solution. Finally, Differential Evolution (DE) is used to determine the optimal solution for fuzzy clustering. The proposed method, namely DE-AFCF, is the first to propose a globally automatic fuzzy CDF algorithm, which not only can automatically determine the number of clusters k but also can search for the optimal fuzzy partition matrix by taking into account both clustering compactness and separation. The DE-AFCF is also applied in some image clustering problems, such as processed image detection, and traffic image recognition. [ABSTRACT FROM AUTHOR]

The dynamic multiobjective evolutionary algorithm (DMOEA) is an efficient solver for dynamic multiobjective optimization problems (DMOPs). It is challenging for algorithms to converge quickly and maintain diversity in new environments. However, existing DMOEAs incorporate strategies only in the environment response stage, which may limit the further improvement of the algorithm performance. To address this problem, different strategies have been proposed in the environment response stage and static optimization stage to balance convergence and diversity throughout the optimization process. In the static optimization stage, nondominated solutions-guided evolution leaves the individuals that are closer to the nondominated individuals among the original individuals and the opposition individuals generated by opposition-based learning, which can accelerate the convergence of each generation. In the environment response stage, based on the individual dominance relationship before environmental change, the fine prediction strategy performs difference prediction and opposition-based learning prediction for nondominated and dominated individuals, respectively, which results in an initial population with good convergence and diversity in the new environment. The performance of proposed algorithm was evaluated on 22 instances and compared to eight state-of-the-art algorithms. The results show that proposed algorithm outperforms its competitors on most problems. Additionally, proposed algorithm performs better in early environmental changes and is relatively insensitive to different severity changes. [ABSTRACT FROM AUTHOR]

In this paper, a transfer learning based evolutionary algorithm (TLEA) framework for multi-objective optimization problems (MOPs) is proposed. In the TLEA framework, a complex multi-objective optimization task is decomposed into a set of relatively simple multi-objective optimization subtasks and then optimized collaboratively by parallel subpopulation searches with the proposed transfer learning method. More specifically, neighboring subtasks may have some similar features during parallel searches of corresponding subpopulations, and those similarities can be exploited through the proposed transfer learning strategy to improve the collaboration among these search subpopulations and achieve greater efficiency. To show the generality of the proposed algorithm framework, two implementations of the proposed TLEA framework based on differential evolution (DE) and particle swarm optimization (PSO), i.e., TLPSO and TLDE, are presented and studied in detail. In TLPSO and TLDE, the subproblem features are reflected by the search subpopulations, which are generated by a pair of specific parameters. Therefore, subpopulations can adaptively adjust parameter settings by learning useful information from neighboring subproblems with more appropriate parameters during the search. The experimental results show that TLPSO performs better than other algorithms on at least five out of 12 test problems in terms of the IGD indicator and on at least seven out of 12 test problems in terms of the HV indicator. TLDE has an advantage over the other algorithms on five out of 12 test problems in terms of the IGD indicator and on seven out of 12 test problems in terms of the HV indicator. [ABSTRACT FROM AUTHOR]

Cross-collection topic models extend previous single-collection topic models, such as Latent Dirichlet Allocation (LDA), to multiple collections. The purpose of cross-collection topic modeling is to model document-topic representations and reveal similarities between each topic and differences among groups. However, the restriction of Dirichlet prior and the significant privacy risk have hampered those models' performance and utility. Training those cross-collection topic models may, in particular, leak sensitive information from the training dataset. To address the two issues mentioned above, we propose a novel model, cross-collection latent Beta-Liouville allocation (ccLBLA), which operates a more powerful prior, Beta-Liouville distribution with a more general covariance structure that enhances topic correlation analysis. To provide privacy protection for the ccLBLA model, we leverage the inherent differential privacy guarantee of the Collapsed Gibbs Sampling (CGS) inference scheme and then propose a hybrid privacy protection algorithm for the ccLBLA model (HPP-ccLBLA) that prevents inferring data from intermediate statistics during the CGS training process without sacrificing its utility. More crucially, our technique is the first attempt to use the cross-collection topic model in image classification applications and investigate the cross-collection topic model's capabilities beyond text analysis. The experimental results for comparative text mining and image classification will show the merits of our proposed approach. [ABSTRACT FROM AUTHOR]

Due to the expanding rate of scientific publications, it has become a necessity to summarize scientific documents to allow researchers to keep track of recent developments. In this paper, we formulate the scientific document summarization problem in a multi-view clustering (MVC) framework. Two views of the scientific documents, semantic and syntactic, are considered jointly in MVC framework. To obtain an improved partitioning corresponding to different views, a differential evolution algorithm is utilized as the underlying optimization strategy. After obtaining the final optimal partitioning, various sentence-level features like length, position of the sentences, among others, are investigated to extract high scoring sentences and these extracted sentences are used to form the summary. We have also investigated the use of (a) two embedding spaces to represent the sentences of the documents in semantic space; and (b) use of citation contexts to incorporate their importance in summarizing a given scientific document. Our proposed multi-view based clustering approach is purely unsupervised in nature and, does not utilize any labelled data for the generation of the summary. To show the potentiality of multi-view clustering, a single-view based clustering framework is also developed for the purpose of comparison. The performance of the multi-view based system in comparison to a single-view is first tested on some random articles selected from the recently released SciSummNet 2019 dataset using ROUGE measure. Then, two more datasets namely, CL-SciSumm 2016 and CL-SciSumm 2017 are considered for evaluation. Our investigations prove that the proposed approach yields better results than the existing supervised and unsupervised methods including transformer-based deep learning model, with statistically significant improvements. [ABSTRACT FROM AUTHOR]

In this paper, a new population initialization method for metaheuristic algorithms is proposed. In our approach, the initial set of candidate solutions is obtained through the sampling of the objective function with the use of the Metropolis–Hastings technique. Under this method, the set of initial solutions adopts a value close to the prominent values of the objective function to be optimized. Different from most of the initialization methods that consider only spatial distribution, in our algorithm, the initial points represent promising regions of the search space, which deserve to be exploited to identify the global optimum. These characteristics allow the proposed approach obtains a faster convergence and improves the quality of the produced solutions. With the objective to demonstrate the capacities of our initialization method, it has been embedded in the classical Differential Evolution algorithm. To evaluate its performance, the complete system has been tested in a set of representative benchmark functions extracted from several datasets. Experimental results show that the proposed technique presents a better convergence speed and an increment in the quality of the solutions than other similar approaches. [ABSTRACT FROM AUTHOR]

The performance of differential evolution (DE) is significantly affected by the selection of mutation strategies and control parameters. Inappropriate selection may lead to premature convergence and stagnation. Therefore, selecting appropriate mutation strategies and control parameters has always been a challenging task. In this paper, a differential evolution with variable leader-adjoint populations (LADE) is proposed. In LADE, a leader-adjoint model is used to divide the population in each generation into leader population and adjoint population. The leader population adopts a novel DE/current-best-rand/1 mutation strategy that can enhance the exploitation ability and avoid evolution stagnation. The adjoint population employs an improved DE/rand/1 mutation strategy that can not only strengthen the exploration ability, but also accelerate individual evolution by guiding the search process to the promising regions. Consequently, the leader-adjoint model can achieve a good balance between exploration and exploitation at different stages of evolution. Moreover, a parameter adaptation method is utilized to dynamically adjust the values of control parameters. To verify the performance of LADE, numerical experiments on the CEC2014 benchmark functions and Lennard-Jones potential real-world problem are executed. Experiment results show that the proposed LADE is significantly better than, or at least comparable to recent and advanced algorithms. In addition, experiments evaluate and analyze the effect of control parameters on the algorithm. [ABSTRACT FROM AUTHOR]

The canonical Pigeon-inspired optimization (PIO) possesses excellent local exploitation capability and can provide a very fast convergence speed, but it is also easily trapped in local optima especially when facing complex problems. In order to take full advantage of the superior local exploitation capability, and at the same time enhance the global exploration capability of PIO, we present a modified PIO called high-level target navigation PIO (HTNPIO). The HTNPIO includes three strategies, selective mutation strategy (SMS), levy-based map-compass strategy (LMS), and enhanced landmark strategy (ELS). In the strategies, two kinds of mutation strategies and a simple levy-flight operator are performed. What's more, an LMS-ELS probability is proposed to balance the exploration and exploitation. In order to test the performance of the proposed optimizer, HTNPIO is made comparisons with other 15 PIO and advanced heuristic algorithms on the IEEE CEC2017 benchmark problems and 5 real world optimization problems. Experimental results demonstrate that HTNPIO defeats all the competitors on the CEC2017 benchmark problems including the extraordinarily competitive LSHADE, and also exhibits extremely competitive performance in dealing with the real-world problems. Therefore, HTNPIO might be effective to provide promising solutions in various function and industrial optimization problems. [ABSTRACT FROM AUTHOR]

In recent years, volatile markets have given rise to a new production system, seru production system, which has been widely utilized in the Asian electronics industry and considered as the next generation of lean production. This paper focuses on the configuration of a seru production system considering that (1) serus are physically reconfigurable rather than fixed and (2) workers have different skill ranges and processing velocities. A mathematical model, whose objectives are to minimize the total completion time and to minimize the total labor cost, is constructed. An improved multiobjective algorithm that combines a multiobjective genetic algorithm, differential evolutionary algorithm and conflict-factor-based mutation operator is proposed in this paper. Computational experiments show that our proposed algorithm outperforms classic scheduling heuristics and well-known algorithms. [ABSTRACT FROM AUTHOR]

This article introduces an efficient surrogate-assisted bi-swarm evolutionary algorithm (SABEA) with hybrid and ensemble strategies for computationally expensive optimization problems. In SABEA, the evolutionary swarm is randomly partitioned into two sub-swarms to maintain diversity of the population. One sub-swarm evolves using the differential evolution (DE) and the other one evolves using teaching-learning-based optimization (TLBO). The proposed SABEA has strong exploration and exploitation capabilities by taking advantages of these two powerful algorithms. Besides, both the global and the local surrogate models cooperate effectively in the proposed SABEA for estimating the fitness value. The global model is established with all samples in the database for global search, and the local model is created with training samples around the current swarm for local search. In addition, a restart mechanism and two model management schemes, namely the individual-based and generation-based, are effectively integrated in the proposed algorithm to make SABEA more strong. Twenty benchmark functions and the tension/compression spring design problem are employed to assess the proposed SABEA. Comprehensive experimental results demonstrate that our proposed SABEA has superior performance comparing with several state-of-the-art competing algorithms on most of the test problems with low, medium and high dimensions. [ABSTRACT FROM AUTHOR]

The deployment of security guards in large public spaces is a promising research topic with a wide range of applications. Existing methods are mainly based on manual design approaches, which are neither effective nor flexible enough for large-scale scenarios. To address this issue, this paper proposes an evolutionary framework to automatically generate the optimal deployment strategy of security guards in large public spaces. The proposed method includes a new metric for automatically evaluating deployment strategies, as well as an evolutionary solver based on differential evolution to optimize the deployment strategy automatically. To evaluate its effectiveness, the proposed evolutionary framework is tested on two synthetic scenarios with different characteristics and one real-world scenario. The results demonstrate that the proposed framework outperforms several commonly used strategies in terms of the response time of security guards. [ABSTRACT FROM AUTHOR]

This work explores the effectiveness and robustness of quantum computing by conjoining the principles of quantum computing with the conventional computational paradigm for the automatic clustering of colour images. In order to develop such a computationally efficient algorithm, two population-based meta-heuristic algorithms, viz., Particle Swarm Optimization (PSO) algorithm and Enhanced Particle Swarm Optimization (EPSO) algorithm have been consolidated with the quantum computing framework to yield the Quantum Inspired Particle Swarm Optimization (QIPSO) algorithm and the Quantum Inspired Enhanced Particle Swarm Optimization (QIEPSO) algorithm, respectively. This paper also presents a comparison between the proposed quantum inspired algorithms with their corresponding classical counterparts and also with three other evolutionary algorithms, viz., Artificial Bee Colony (ABC), Differential Evolution (DE) and Covariance Matrix Adaption Evolution Strategies (CMA-ES). In this paper, twenty different sized colour images have been used for conducting the experiments. Among these twenty images, ten are Berkeley images and ten are real life colour images. Three cluster validity indices, viz., PBM, CS-Measure (CSM) and Dunn index (DI) have been used as objective functions for measuring the effectiveness of clustering. In addition, in order to improve the performance of the proposed algorithms, some participating parameters have been adjusted using the Sobol's sensitivity analysis test. Four segmentation evaluation metrics have been used for quantitative evaluation of the proposed algorithms. The effectiveness and efficiency of the proposed quantum inspired algorithms have been established over their conventional counterparts and the three other competitive algorithms with regards to optimal computational time, convergence rate and robustness. [ABSTRACT FROM AUTHOR]

During recent decades, multi-objective optimization has aroused extensive attention, and a variety of related algorithms have been proposed. A hybrid multi-objective optimization algorithm based on angle competition and neighborhood protection mechanism (HCPMOEA) is proposed in this paper. First, an environmental selection strategy based on neighborhood protection is introduced to make great compromises between optimization performance and time consumption. Then, the difference between Genetic algorithm and Differential evolution is analyzed from the perspective of offspring distribution and a hybrid operator is proposed to obtain good balances between exploration and exploitation. Besides, an elite set is employed to improve chances of the superior solutions generating offspring, and angle competition strategy is adopted to realize optimization matching of parents, thus improving the quality of offspring. The performance of HCPMOEA has been proved by comparing with 13 classic or state-of-the-arts algorithms on 19 standard benchmark, and the corresponding results show the competitive advantages in effectiveness and efficiency. In addition, the practicality of the proposed HCPMOEA is further verified by two real-world instances. Therefore, all of the aforementioned results have proved the superiority of the proposed HCPMOEA in solving bi-objective and tri-objective problems. [ABSTRACT FROM AUTHOR]

Trapping in local solutions is the main issue in several metaheuristic techniques. To solve such drawbacks by enhancing the search agents, a modified search strategy becomes a more attractive tactic. In this paper, an innovative version of Manta Ray Foraging Optimization (MRFO) is proposed to solve its crucial drawbacks while handling global and engineering optimization problems. The proposed version presents an integrated variant of MRFO with the triangular mutation operator and orthogonal learning strategy, called MRTMO. The two approaches are considered to achieve a robust equipoise between algorithm cores and provide a reliable mechanism to guide the search agents during the optimization process. The proposed MRTMO was tested with challenging CEC2005 and CEC2017 functions and six engineering problems to show its performance. Additionally, several evaluation metrics were employed to ensure the efficiency and robustness of the proposed MRTMO. Furthermore, extensive comparisons with existing optimization algorithms were carried out to ensure the superiority of MRTMO. The numerical experiments proved the competitive performance of the proposed MRTMO in solving all tested CEC optimization and engineering problems. [ABSTRACT FROM AUTHOR]

Dissolved oxygen is a critical water quality factor, and its prediction and control have long been a research hotspot in the aquaculture industry. This study aims to address the limitations of poor population diversity and the ease of falling into local optima in the basic sparrow search algorithm (SSA). Herein, we improved this methodology in many aspects and proposed a hybrid improved SSA (HISSA) to enhance optimization performance. First, in the global search process of the sparrow producer, the position update strategy is modified to enhance the global search ability of the algorithm. Second, a butterfly algorithm is incorporated to replace the position update strategy in the escape stage of the producer to promote information exchange between individuals. Third, the Cauchy mutation strategy and greedy rule are used to perturb and update the current optimal solution to improve the algorithm's ability to jump out of local optima. The performance of the proposed HISSA was verified on 10 benchmark functions. The experimental results illustrate that the proposed algorithm outperformed the four intelligent algorithms evaluated herein. In addition, the proposed HISSA was separately applied to the parameter optimization of the two most commonly used engineering methods: the back propagation neural network (BPNN) and proportional-integral-derivative (PID) controller. Compared with other intelligent algorithm-optimized methods, the error evaluation index values of the HISSA-BPNN and HISSA-PID controllers are small, indicating that the proposed HISSA is highly effective and practical for engineering parameter optimization, and can be competent for the prediction and control of dissolved oxygen. [ABSTRACT FROM AUTHOR]

Accurate prediction of oil consumption plays a dominant role in oil supply chain management. However, because of the effects of the coronavirus disease 2019 (COVID-19) pandemic, oil consumption has exhibited an uncertain and volatile trend, which leads to a huge challenge to accurate predictions. The rapid development of the Internet provides countless online information (e.g., online news) that can benefit predict oil consumption. This study adopts a novel news-based oil consumption prediction methodology–convolutional neural network (CNN) to fetch online news information automatically, thereby illustrating the contribution of text features for oil consumption prediction. This study also proposes a new approach called attention-based JADE-IndRNN that combines adaptive differential evolution (adaptive differential evolution with optional external archive, JADE) with an attention-based independent recurrent neural network (IndRNN) to forecast monthly oil consumption. Experimental results further indicate that the proposed news-based oil consumption prediction methodology improves on the traditional techniques without online oil news significantly, as the news might contain some explanations of the relevant confinement or reopen policies during the COVID-19 period. [ABSTRACT FROM AUTHOR]

The exploration ability and exploitation ability of natural computing methods are two abilities that restrict each other. In order to balance the search breadth of the algorithm to the target space and the convergence accuracy of the algorithm, and improve the effect of the algorithm, this paper proposes a natural computing method of multi-spatially cooperative game based on clustering, and analyzes its convergence. Based on the idea of population space grouping, the algorithm performs the game calculation of fusion cosine similarity reverse strategy for particles in the same space, and performs binary crossover and polynomial mutation. In this paper, the algorithm is compared with the classical algorithm and the excellent algorithm in recent years on 12 classical test functions, and the 28 standard functions of CEC2013 are used for experimental evaluation. Finally, the algorithm is applied to the split torque transmission optimization of the main reducer. The experimental results show that the multi-space division of the population based on clustering is conducive to the population jumping out of the local extreme value, and the game calculation integrating cosine similarity is conducive to improving the convergence accuracy of the population. The practical engineering application also proves that the method has good universality and effectiveness. [ABSTRACT FROM AUTHOR]

Multifactorial evolution algorithm (MFEA) is a powerful search paradigm with the purpose of addressing multiple optimization tasks simultaneously in the field of evolutionary computation. The assortative mating of MFEA is a key component to make it outperform traditional single-task optimization algorithms. However, the optimal solution of each generation has still not been well utilized to accelerate convergence in the process of assortative mating operation for existing MFEAs. This paper proposes a multifactorial differential evolution algorithm with optima-based transformation (MFDE-OBT), which employs the optimal solution of each generation to design an improved assortative mating operation based on the DE/rand/2 mutation. The improved operation generates an offspring for each individual by running a perturbation around a current optimal individual. The object of the assortative mating is a vector added for the perturbation, which is the sum of two difference vectors generated by a random sample with a certain probability from an individual set that can be same as or different from the one of the task involving the optimal individual. In addition, MFDE-OBT integrates an opposition-based search strategy behind the assortative mating operation to balance exploitation and exploration of each search space. Experimental results on benchmark problems constituted by tasks with different degree of similarity and intersection demonstrate the advantage of the proposed MFDE-OBT algorithm over some state-of-the-art algorithms, in terms of solution precision and convergence performance. [ABSTRACT FROM AUTHOR]

The short-time aggregation of human traffic places high demands on the communication capacity of cellular networks. The deployment of expensive permanent infrastructure without continuous high traffic is uneconomical, and the problem poses a challenge. In this study, a green mall traffic model based on mobile base stations with a dynamic sleep strategy is proposed for surges of shopping mall traffic. The model is addressed through a modified improved differential evolution (MIDE) algorithm based on the original improved differential evolution (IDE) algorithm. The algorithm has two sets of mutation and restart policies adapted to different traffic volumes, and can dynamically adjust according to the traffic volume. The effectiveness of the algorithm is verified by simulation experiments. Compared with the traditional differential evolution (DE) algorithm and the DE series algorithms recently published in Swarm and Evolutionary Computation, a journal with high impact factors, MIDE can effectively optimize the system model and improve its energy efficiency, saving 1.1%–56.4% in simulation experiments. [ABSTRACT FROM AUTHOR]

The Salp Swarm Algorithm (SSA) is an effective single-objective optimization algorithm that was inspired by the navigating and foraging behaviors of salps in their natural habitats. Although SSA was successfully tailored and applied to solve various types of optimization problems, it often suffers from premature convergence and typically does not perform well with high-dimensional optimization problems. This paper introduces an Improved SSA (ISSA) algorithm to enhance the performance of SSA in solving single-objective continuous optimization problems. ISSA has four characteristics. First, it employs Gaussian Perturbation to improve the diversity of initial population. Second, it uses highly disruptive polynomial mutation (HDPM) to update the leader salp in the salp chain. Third, it uses the Laplace crossover operator to improve its exploration ability. Fourth, it uses a new opposition learning method called Mixed Opposition-based Learning (MOBL) to improve its convergence rate and exploration ability. A set of 14 standard benchmark functions was used to evaluate the performance of ISSA and compare it to three variations of SSA (SSA, Hybrid SSA with Particle Swarm Optimization HSSAPSO Singh et al. (2020) and Enhanced SSA (ESSA) Zhang et al. (2020)). The overall experimental and statistical results indicate that ISSA is a better optimization algorithm than the other SSA variations. Further, the single-objective IEEE CEC 2014 (IEEE Congress on Evolutionary Computation 2014) functions were used to evaluate and compare the performance of ISSA to 18 well-known and state-of-the-art optimization algorithms (Exploratory Cuckoo Search (ECS) Abed-alguni (2021)), Grey Wolf Optimizer (GWO) Mirjalili and Mirjalili (Advances in Engineering Software, 69, 46–61, 2014), Distributed Grey Wolf Optimize (DGWO) Abed-alguni and Barhoush (2018), Cuckoo Search (CS) Yang and Deb (2009), Distributed adaptive differential evolution with linear population size reduction evolution (L-SHADE) Tanabe and Fukunaga (2014), Memory-based Hybrid Dragonfly Algorithm (MHDA) KS and Murugan (Expert Syst Appl, 83, 63–78, 2017), Fireworks Algorithm with Differential Mutation (FWA-DM) Yu et al. (2014), Differential Evolution-based Salp Swarm Algorithm (DESSA) Dhabal et al. (Soft Comput, 25(3), 1941–1961, 2021), LSHADE with Fitness and Diversity Ranking-Based Mutation Operator (FD-LSHADE) Cheng et al. (Swarm and Evolutionary Computation, 61, 100816, 2021), Distance based SHADE (Db-SHADE) Viktorin et al. (Swarm and Evolutionary Computation, 50, 100462, 2019) and Zeng et al. (Knowl-Based Syst, 226, 107150, 2021), Mean–Variance Mapping Optimization (MVMO) Iacca et al. (Expert Syst Appl, 165, 113902, 2021), Time-varying strategy-based Differential Evolution (TVDE) Sun et al. (Soft Comput, 24(4), 2727–2747, 2020), Butterfly Optimization Algorithm with adaptive gbest-guided search strategy and Pinhole-Imaging-based Learning (PIL-BOA)Long et al. (Appl Soft Comput, 103, 107146, 2021), Memory Guided Sine Cosine Algorithm (MG-SCA) Gupta et al. (Eng Appl Artif Intell, 93, 103718, 2020), Lévy flight Jaya Algorithm (LJA) Iacca et al. (2021), Sine Cosine Algorithm (SCA) Dhabala et al. (2021), Covariance Matrix Adaptation Evolution Strategy (CMA-ES) Hansen et al. (Evolutionary Computation, 11(1), 1–18, 2003) and Coyote Optimization Algorithm (COA) Pierezan and Coelho (2018)). The results indicate that ISSA performs better than the tested optimization algorithms. [ABSTRACT FROM AUTHOR]

Image registration is the process of matching and superimposing multiple images from heterogeneous sources, and it is widely used in areas such as image processing and computer vision. In point feature-based image registration, the RANdom SAmple Consensus (RANSAC) algorithm is commonly used to eliminate mis-matches and solve transform models. Traditional RANSAC tends to sample in sample space. Its discrete estimation of model parameters can lead to an inability to solve for the largest consensus set at large input sizes and low inlier ratio. RANSAC with sampling in parameter space is a novel and effective method, but it is less studied and still has many shortcomings. In this paper, a Differential Evolution SAmple Consensus (DESAC) algorithm based on sampling in parameter space is proposed, which represents the model as individuals and optimizes the model parameters by Differential Evolution (DE). First, for an adequate search of the parameter space, DESAC uses priori information for the first initialization and the resident initialization performed in the iteration. Second, to achieve a balance between exploration and exploitation, a mutation operator combining neighborhood optimal individuals and random individuals is proposed. Third, a novel selection operator is proposed in order to obtain individuals with high number of inliers and low inlier error. Finally, to reduce the verification overhead, a simplified pre-test step is applied before the full verification. Comparing with advanced RANSAC variants in real image dataset, the proposed method can robustly return consensus sets with very high number of inliers. Its average inlier error is lower and the transformation model is more accurate. [ABSTRACT FROM AUTHOR]

Prediction-error expansion (PEE) is an important technique for reversible watermarking (RW). Traditional PEE methods exploit pixel correlations by modifying prediction-error histogram (PEH) with Laplacian-like distributions, but they rarely consider the influence of needless shifting on image distortion. Asymmetric PEE approaches utilize skewed PEHs to select fewer shifted pixels for superior performance in capacity-distortion control tasks. In this way, the construction of a suitable extreme predictor is especially useful for increasing embedding capacity and reducing image distortion to satisfy user requirements. However, in recent works, the parameters of extreme predictors were not fully optimized. This paper presents a novel extreme predictor, in which the prediction parameters are optimized by using a differential evolution (DE) algorithm to increase the performance of asymmetric PEE by building a sharp and skewed PEH. Specifically, we develop a new DE algorithm based on hybrid mutation and grouping crossover operations to further improve the optimization accuracy of the algorithm. Then, an RW scheme based on the proposed predictor and image separation is built to fully exploit the embedding potential of different bit-planes in images. To reduce image distortion, a strategy that is complementary to double-embedding is presented for lower bit-plane (LBP) embedding. Experimental results demonstrate that the proposed scheme can achieve better image quality and a larger embedding capacity than similar state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Favourable performance of designed controllers for Permanent magnet synchronous motors (PMSMs), deeply depends on accurate model and parameters of PMSM. This paper proposes two improved versions of particle swarm optimization (PSO) for identification of all-six electrical and mechanical parameters of PMSM. This research inserts two different strategies to overcome premature convergence of PSO. In the first version, the PSO is incorporated with quasi-opposition-based learning (QOBL) to be accelerated and also to diversify its search moves. In the second version of proposed improved PSO, in an attempt to diversify and manifold search moves, a chaotic local search is inserted in the PSO to further enhance its global search ability. Aforementioned algorithms are tested on problem of PMSM parameter identification and also 30 CEC2014 benchmark functions. The obtained results demonstrate that the proposed algorithms in this research beside of good solution quality are very effective and robust so that they produce similar and promising results over repeated runs. Subsequently, a comparison of the proposed algorithm with two recent well performing algorithms, i.e. covariance matrix adaptation-evolution strategy (CMA-ES) and success-history based adaptive differential evolution with linear population size reduction (L-SHADE) confirmed a comparable performance of our proposed algorithm. Statistical analysis of obtained results on CEC2014 functions by Wilcoxon test also indicated that the proposed algorithm has a significant performance over other compared state-of-art algorithms. [ABSTRACT FROM AUTHOR]

As one of the main contents of influence analysis, influence maximization is selecting a group of influential nodes with specified size in a given network to form a seed node set, and the influence spread cascaded by the selected seed node set can be maximized under a given propagation model. The research of influence maximization is helpful to understand social network and viral marketing. How to develop an effective algorithm to solve this problem in large-scale networks is still a challenge. In this paper, a discrete differential evolution algorithm based on community structure (CDDE) is proposed. At first, the fast Louvain algorithm is used to detect the community structure. On this basis, significant communities are defined and candidate nodes are extracted from each significant community. And then, an improved discrete differential evolution algorithm is proposed to obtain influential nodes. Furthermore, a population initialization strategy based on candidate nodes is designed, and the candidate nodes are also used to accelerate the discrete evolution process of the population. Experimental results on six real-world social networks show that the proposed CDDE is competitive with the comparison algorithms in terms of effectiveness and efficiency, and achieves comparable influence spread to CELF. [ABSTRACT FROM AUTHOR]

To solve the problem of current popular clustering algorithms needing to set the number of clusters and hyperparameters according to prior knowledge, we use the average nearest neighbour distance, a statistic that represents the characteristics of sample aggregation in the data space, and propose a two-stage clustering algorithm based on evolution and propagation patterns (EPC). In the evolution stage, the EPC algorithm obtains the initial clustering results and the number of clusters by evolving a small number of samples from random sampling in the data space in an incremental way. According to the nearest neighbour principle, the EPC propagates the cluster labels of the initial clustering results to the unlabelled samples in the propagation stage. Furthermore, the EPC algorithm uses a correction mechanism. It adopts Monte Carlo multiple simulation methods in the evolution stage to improve the stability of clustering results obtained by random sampling. Experiments on datasets and applications on image segmentation datasets show that the EPC algorithm is superior to the current popular clustering algorithm in performance. Finally, we conducted a systematic and comprehensive analysis of the EPC algorithm through ablation experiments, showing that the EPC algorithm has good robustness. [ABSTRACT FROM AUTHOR]

An efficient survey of numerous traditional metaheuristic algorithms (MAs) has been investigated in this paper. Among successful MAs, differential evolution (DE) and particle swarm optimization (PSO) have been widely recognized to solve complex optimization problems and received much attention from many researchers. Therefore, DE and PSO are chosen in the present study and an extensive survey of their recent-past variants with hybrids has been inspected again. After this an advanced DE (ADE) and PSO (APSO) with their hybrid (AHDEPSO) are proposed for unconstrained optimization problems. In ADE a novel mutation strategy, crossover probability and random nature selection scheme (to avoid premature convergence) as well as in APSO novel gradually varying parameters (to avoid stagnation) are introduced. Hence, ADE and APSO affords different convergence characteristics to the solution space. Also to balance between exploration and exploitation, in AHDEPSO population is divided (multi-population approach) and merged with others in a pre-defined way. Thus, AHDEPSO achieves better solutions and it is expected to obtain productive solutions with an increasing success rate at each cycle. To verify the performance of all 3 proposed algorithms i.e. ADE, APSO, and AHDEPSO applied to solve 23 basic, 30 IEEE CEC 2017 unconstrained benchmark functions and 3 real-world problems. There are several numerical and graphical analyses have been done to verify the performances of the proposed algorithms robustly. Additionally, statistical and comparative analysis confirms the superiority of the proposed algorithms among traditional DE and PSO with their recent variants and hybrids as well as over many state-of-the-art algorithms. Finally, between 3 proposed algorithms the best one i.e. AHDEPSO is recommended to solve unconstrained optimization problems. [ABSTRACT FROM AUTHOR]

The performance of evolutionary algorithms (EAs) is strongly affected by their configurations. Thus, algorithm configuration (AC) problem, that is, to properly set algorithm's configuration, including the operators and parameter values for maximizing the algorithm's performance on given problem(s) is an essential and challenging task in the design and application of EAs. In this paper, an online algorithm configuration (OAC) approach is proposed for differential evolution (DE) algorithm to adapt its configuration in a data-driven way. In our proposed OAC, the multi-armed bandit algorithm is adopted to select trial vector generation strategies for DE, and the kernel density estimation method is used to adapt the associated control parameters during the evolutionary search process. The performance of DE algorithm using the proposed OAC (OAC-DE) is evaluated on a benchmark set of 30 bound-constrained numerical optimization problems and compared with several adaptive DE variants. Besides, the influence of OAC's hyper-parameter on its performance is analyzed. The comparison results show OAC-DE achieves better average performance than the compared algorithms, which validates the effectiveness of the proposed OAC. The sensitivity analysis indicates that the hyper-parameter of OAC has little impact on OAC-DE's performance. [ABSTRACT FROM AUTHOR]

The present paper proposes a new algorithm designed for solving optimization problems. This algorithm is a hybrid of Differential Evolution (DE) and Particle Swarm Optimization (PSO) algorithms. The proposed algorithm uses a coalition or cooperation model in the game theory to combine the DE and PSO algorithms. This is done in an attempt to keep a balance between the exploration and exploitation capabilities by preventing population stagnation and avoiding the local optimum. The DE and PSO algorithms are two players in the state space, which play cooperative games together using the Nash bargaining theory to find the best solution. To evaluate the performance of the proposed algorithm, 25 benchmark functions are used in terms of the CEC2005 structure. The proposed algorithm is then compared with the classical DE and PSO algorithms and the hybrid algorithms recently proposed. The results indicated that the proposed hybrid algorithm outperformed the classical algorithms and other hybrid models. [ABSTRACT FROM AUTHOR]

This paper presents a backtracking differential evolution with multi-mutation strategies autonomy and collaboration (bDE-MsAC) to solve the optimization problems. In the proposed bDE-MsAC, five modified mutation strategies are employed to simultaneously construct a global exploration domain (GED) and a local exploitation domain (LED). Then, a mechanism of multi-mutation strategies autonomy and collaboration is introduced to realize the coevolution between GED and LED. Besides, the parameter adaptation scheme based on individual similarity and evolution status can adaptively update the parameters and bring vitality to the evolution process. Meanwhile, an evolution backtracking strategy is designed to control the population diversity. The population can trace back to the generation with maximum best fitness descent and then change the search direction to avoid the premature. Comparison results with nine DE algorithms on the well-known test functions reveal that the proposed bDE-MsAC has a competitive performance in comparison with other DE methods. In addition, the experiments analyze the effect of two key parameters and demonstrate the effectiveness and superiority of the evolution backtracking strategy. [ABSTRACT FROM AUTHOR]

Influence nets (INs) are proposed on the basis of causal logic for the purpose of depicting causal relationship strengths in complex systems. However, it is difficult to accurately determine the causal strength (CAST) parameters on the basis of expert knowledge only, particularly for practical problems with multiple attributes and multiple states. In this paper, the original IN with binary states is first extended into a multi-state IN, endowing the IN with the ability to model and infer under multiple states. Based on this method, an optimization framework is proposed to optimize the CAST parameters of multi-state INs. Finally, two practical cases are studied to verify the feasibility and efficiency of the proposed multi-state IN with an optimization framework in regression and classification. [ABSTRACT FROM AUTHOR]

The exponential growth in the number of scientific articles has made it difficult for the researchers to keep themselves updated with the new developments. Scientific document summarization solves this problem by providing a summary of essential contributions. In this paper, we have presented a novel method of scientific document summarization using a multi-objective differential evolution technique. Here, firstly distinct important sentences are extracted by using citation contextualization. These sentences are further clustered using the concept of multi-objective clustering. Two objective functions, PBM index, and XB index, measuring the compactness and separation of sentence clusters, are simultaneously optimized utilizing the search capability of multi-objective differential evolution. We have conducted our experiments on CL-SciSumm 2016, CL-SciSumm 2017, CL-SciSumm 2018, and CL-SciSumm 2019 datasets. Obtained results of CL-SciSumm 2016 and CL-SciSumm 2017 are compared with the state-of-the-art methods. Evaluation results demonstrate that our method outperforms others in terms of ROUGE-2, ROUGE-3, and ROUGE-SU4 scores. [ABSTRACT FROM AUTHOR]

Social media platforms become paramount for gathering relevant information during the occurrence of any natural disaster. Twitter has emerged as a platform which is heavily used for the purpose of communication during disaster events. Therefore, it becomes necessary to design a technique which can summarize the relevant tweets and thus, can help in the decision-making process of disaster management authority. In this paper, the problem of summarizing the relevant tweets is posed as an optimization problem where a subset of tweets is selected using the search capability of multi-objective binary differential evolution (MOBDE) by optimizing different perspectives of the summary. MOBDE deals with a set of solutions in its population, and each solution encodes a subset of tweets. Three versions of the proposed approach, namely, MOOTS1, MOOTS2, and MOOTS3, are developed in this paper. They differ in the way of working and the adaptive selection of parameters. Recently developed self-organizing map based genetic operator is explored in the optimization process. Two measures capturing the similarity/dissimilarity between tweets, word mover distance and BM25 are explored in the optimization process. The proposed approaches are evaluated on four datasets related to disaster events containing only relevant tweets. It has been observed that all versions of the developed MOBDE framework outperform the state-of-the-art (SOA) techniques. In terms of improvements, our best-proposed approach (MOOST3) improves by 8.5% and 3.1% in terms of ROUGE− 2 and ROUGE−L, respectively, over the existing techniques and these improvements are further validated using statistical significance t-test. [ABSTRACT FROM AUTHOR]

In machine learning, searching for the optimal feature subset from the original datasets is a very challenging and prominent task. The metaheuristic algorithms are used in finding out the relevant, important features, that enhance the classification accuracy and save the resource time. Most of the algorithms have shown excellent performance in solving feature selection problems. A recently developed metaheuristic algorithm, gaining-sharing knowledge-based optimization algorithm (GSK), is considered for finding out the optimal feature subset. GSK algorithm was proposed over continuous search space; therefore, a total of eight S-shaped and V-shaped transfer functions are employed to solve the problems into binary search space. Additionally, a population reduction scheme is also employed with the transfer functions to enhance the performance of proposed approaches. It explores the search space efficiently and deletes the worst solutions from the search space, due to the updation of population size in every iteration. The proposed approaches are tested over twenty-one benchmark datasets from UCI repository. The obtained results are compared with state-of-the-art metaheuristic algorithms including binary differential evolution algorithm, binary particle swarm optimization, binary bat algorithm, binary grey wolf optimizer, binary ant lion optimizer, binary dragonfly algorithm, binary salp swarm algorithm. Among eight transfer functions, V4 transfer function with population reduction on binary GSK algorithm outperforms other optimizers in terms of accuracy, fitness values and the minimal number of features. To investigate the results statistically, two non-parametric statistical tests are conducted that concludes the superiority of the proposed approach. [ABSTRACT FROM AUTHOR]

In the multimodal multi-objective optimization problems (MMOPs), at least two equivalent Pareto optimal solutions in decision space with an identical objective value are desired. The challenge for solving MMOPs is locating equivalent Pareto optimal solutions in decision space, and maintaining a fine balance between diversity and convergence of Pareto optimal solutions in both decision space and objective space, simultaneously. To address this issue, a success-history based parameter adaptation for multimodal multi-objective differential evolution algorithm using fitness sharing (MMOSHADE) is proposed in this paper. A success-history based parameter adaptation for differential evolution (SHADE) is integrated into MMOSHADE to find elite individuals and locate Pareto optimal solutions in decision space. Subsequently, a modified selection operation in differential evolution (DE) is introduced into MMOSHADE to explore outstanding convergence solutions. Furthermore, a double fitness sharing method is available for maintaining the diversity of Pareto optimal solutions in both decision space and objective space, simultaneously. The proposed MMOSHADE is performed on three categories of problems to test the performance of MMOSHADE. The comparison between MMOSHADE and six competing algorithms demonstrates the superiority of the proposed MMOSHADE in solving MMOPs and large-scale polygon-based MMOPs. MMOSHADE is also capable of finding the entire Pareto front in most cases when it is used to address multi-objective optimization problems. Additionally, the effectiveness of several strategies is validated by the designed experiments, and the parameters involved in MMOSHADE are discussed. [ABSTRACT FROM AUTHOR]

The main aim of multimodal optimization problems (MMOPs) is to find and deal with multiple optimal solutions using an objective function. MMOPs perform the exploration and exploitation simultaneously in the search space. The novelty of this paper includes the following improvements in differential evolution to be able to solve MMOPs. Clusters are formed from the whole population by applying a niching technique which uses the softmax strategy to assign a cutting probability to the species. Then iterative mutation strategy is followed to generate the unbiased mutant vector. Further, Multi-Armed Bandit (MAB) strategy is used to ensure that new individuals are generated in promising areas. The experimentation of the proposed algorithm has been performed on 20 benchmark functions from IEEE Congress on Evolutionary Computation 2013 (CEC2013). The results depict that the proposed algorithm can be compared with 15 state-of-the-art multimodal optimization algorithms in terms of locating accurate optimal solutions. [ABSTRACT FROM AUTHOR]

This paper presents a novel differential evolution (DE) framework with a self-regulated neighborhood, termed self-regulated differential evolution (SrDE), for real parameter optimization. The novelty and advantage of SrDE are to present a self-regulated neighborhood (SrN) for learning, regulating, and using the neighborhood information of the population in guiding the search process of DE. Specifically, SrDE is characterized by the following three aspects. First, a one-dimensional self-organizing map (SOM) method is employed to dynamically construct the neighborhood relationships between individuals. Second, a self-sizing technique is applied to adaptively regulate the neighborhood size of each individual based on its search state. Third, a neighborhood path-assisted strategy is proposed to utilize promising neighborhood information for guiding the mutation process. Extensive experiments on a suite of real-parameter functions and real-world problems have demonstrated the superior and competitive performance of SrDE when compared with the state-of-the-art DE variants. [ABSTRACT FROM AUTHOR]