Coevolutionary Algorithm Research Articles

A Large-Scale Combinatorial Many-Objective Evolutionary Algorithm for Intensity-Modulated Radiotherapy Planning

Intensity-modulated radiotherapy (IMRT) is one of the most popular techniques for cancer treatment. However, existing IMRT planning methods can only generate one solution at a time and, consequently, medical physicists should perform the planning process many times to obtain diverse solutions to meet the requirement of a clinical case. Meanwhile, multiobjective evolutionary algorithms (MOEAs) have not been fully exploited in IMRT planning since they are ineffective in optimizing the large number of discrete variables of IMRT. To bridge the gap, this article formulates IMRT planning into a large-scale combinatorial many-objective optimization problem and proposes a coevolutionary algorithm to solve it. In contrast to the existing MOEAs handling high-dimensional search spaces via variable grouping or dimensionality reduction, the proposed algorithm evolves one population with fine encoding for local exploitation and evolves another population with rough encoding for global exploration. Moreover, the convergence speed is further accelerated by two customized local search strategies. The experimental results verify that the proposed algorithm outperforms state-of-the-art MOEAs and IMRT planning methods on a variety of clinical cases.

Merged Differential Grouping for Large-Scale Global Optimization

The divide-and-conquer strategy has been widely used in cooperative co-evolutionary algorithms to deal with large-scale global optimization problems, where a target problem is decomposed into a set of lower-dimensional and tractable subproblems to reduce the problem complexity. However, such a strategy usually demands a large number of function evaluations to obtain an accurate variable grouping. To address this issue, a merged differential grouping (MDG) method is proposed in this article based on the subset–subset interaction and binary search. In the proposed method, each variable is first identified as either a separable variable or a nonseparable variable. Afterward, all separable variables are put into the same subset, and the nonseparable variables are divided into multiple subsets using a binary-tree-based iterative merging method. With the proposed algorithm, the computational complexity of interaction detection is reduced to <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$O(\max \{n,n_{ns}\times \log _{2} k\})$ </tex-math></inline-formula> , where <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$n$ </tex-math></inline-formula> , <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$n_{ns}(\leq n)$ </tex-math></inline-formula> , and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$k( &lt; n)$ </tex-math></inline-formula> indicate the numbers of decision variables, nonseparable variables, and subsets of nonseparable variables, respectively. The experimental results on benchmark problems show that MDG is very competitive with the other state-of-the-art methods in terms of efficiency and accuracy of problem decomposition.

A Novel Self-Adaptive Cooperative Coevolution Algorithm for Solving Continuous Large-Scale Global Optimization Problems

Unconstrained continuous large-scale global optimization (LSGO) is still a challenging task for a wide range of modern metaheuristic approaches. A cooperative coevolution approach is a good tool for increasing the performance of an evolutionary algorithm in solving high-dimensional optimization problems. However, the performance of cooperative coevolution approaches for LSGO depends significantly on the problem decomposition, namely, on the number of subcomponents and on how variables are grouped in these subcomponents. Also, the choice of the population size is still an open question for population-based algorithms. This paper discusses a method for selecting the number of subcomponents and the population size during the optimization process (“on fly”) from a predefined pool of parameters. The selection of the parameters is based on their performance in the previous optimization steps. The main goal of the study is the improvement of coevolutionary decomposition-based algorithms for solving LSGO problems. In this paper, we propose a novel self-adapt evolutionary algorithm for solving continuous LSGO problems. We have tested this algorithm on 15 optimization problems from the IEEE LSGO CEC’2013 benchmark suite. The proposed approach, on average, outperforms cooperative coevolution algorithms with a static number of subcomponents and a static number of individuals.

An efficient multi-robot path planning solution using A* and coevolutionary algorithms

Multi-robot path planning has evolved from research to real applications in warehouses and other domains; the knowledge on this topic is reflected in the large amount of related research published in recent years on international journals. The main focus of existing research relates to the generation of efficient routes, relying the collision detection to the local sensory system and creating a solution based on local search methods. This approach implies the robots having a good sensory system and also the computation capabilities to take decisions on the fly. In some controlled environments, such as virtual labs or industrial plants, these restrictions overtake the actual needs as simpler robots are sufficient. Therefore, the multi-robot path planning must solve the collisions beforehand. This study focuses on the generation of efficient collision-free multi-robot path planning solutions for such controlled environments, extending our previous research. The proposal combines the optimization capabilities of the A* algorithm with the search capabilities of co-evolutionary algorithms. The outcome is a set of routes, either from A* or from the co-evolutionary process, that are collision-free; this set is generated in real-time and makes its implementation on edge-computing devices feasible. Although further research is needed to reduce the computational time, the computational experiments performed in this study confirm a good performance of the proposed approach in solving complex cases where well-known alternatives, such as M* or WHCA, fail in finding suitable solutions.

On Improving Adaptive Problem Decomposition Using Differential Evolution for Large-Scale Optimization Problems

Modern computational mathematics and informatics for Digital Environments deal with the high dimensionality when designing and optimizing models for various real-world phenomena. Large-scale global black-box optimization (LSGO) is still a hard problem for search metaheuristics, including bio-inspired algorithms. Such optimization problems are usually extremely multi-modal, and require significant computing resources for discovering and converging to the global optimum. The majority of state-of-the-art LSGO algorithms are based on problem decomposition with the cooperative co-evolution (CC) approach, which divides the search space into a set of lower dimensional subspaces (or subcomponents), which are expected to be easier to explore independently by an optimization algorithm. The question of the choice of the decomposition method remains open, and an adaptive decomposition looks more promising. As we can see from the most recent LSGO competitions, winner-approaches are focused on modifying advanced DE algorithms through integrating them with local search techniques. In this study, an approach that combines multiple ideas from state-of-the-art algorithms and implements Coordination of Self-adaptive Cooperative Co-evolution algorithms with Local Search (COSACC-LS1) is proposed. The self-adaptation method tunes both the structure of the complete approach and the parameters of each algorithm in the cooperation. The performance of COSACC-LS1 has been investigated using the CEC LSGO 2013 benchmark and the experimental results has been compared with leading LSGO approaches. The main contribution of the study is a new self-adaptive approach that is preferable for solving hard real-world problems because it is not overfitted with the LSGO benchmark due to self-adaptation during the search process instead of a manual benchmark-specific fine-tuning.

A dual-population based bidirectional coevolution algorithm for constrained multi-objective optimization problems

The balance between multiple objectives and various constraints is the key to solving constrained multi-objective optimization problems (CMOPs). When dealing with CMOPs with complex feasible regions, some evolutionary algorithms suffer from great challenges in converging to the constrained Pareto front (CPF) with well-distributed feasible solutions. To address this issue, this paper proposes a dual-population based bidirectional coevolution algorithm, called DBC-CMOEA, which aims to converge to the CPF using promising solutions explored from both feasible and infeasible regions. To do so, DBC-CMOEA maintains two populations and an archive, where the dual-population is complementary in the search process and the archive is used to retain promising feasible and infeasible solutions, thus facilitating information exchange between these two populations. For updating the archive, a nondominated sorting procedure and an angle-based selected scheme are conducted to store infeasible and feasible solutions, as they can help to maintain the diversity of the search and find more feasible regions. To evolve the CPF from the bidirectional side of the feasible region, a novel mating selection strategy is used to choose appropriate mating parents. In comparison with some related constraint multi-objective optimization algorithms on a number of benchmark problems, experimental results show that the proposed algorithm performs better than the state-of-the-art constrained multi-objective evolutionary optimizers.

Reducing tardy batches by seru production: Model, exact solution, cooperative coevolution solution, and insights

Seru production, as an innovative production mode, has the advantages of quick response, good flexibility, and high efficiency. We investigate how to reduce tardy batches using seru production. The model of seru production, which minimizes the number of tardy batches, is formulated and decomposed into seru formation and seru scheduling. To obtain the optimal solution, we formulate a linear model of seru scheduling to maximize the number of on-time batches for a given seru formation. In the case of small-scale instances, the optimal solution for seru production is obtained by solving the linear model exactly for all seru formations. However, this method is extremely time-consuming. Therefore, for medium-scale instances, we develop an exact method based on surrogate-relaxation and branch-and-bound. Surrogate relaxation is developed to rapidly eliminate unpromising seru formations, where an upper bound of a given seru formation is quickly produced. If the upper bound is worse than the current lower bound of seru production, the seru formation is unpromising and eliminated in advance. Otherwise, it remains, and its feasible solution is obtained by surrogate relaxation to update the current lower bound. By the proposed method, 97.29% of unpromising seru formations are eliminated on average and 76.21% of computational time is saved. Subsequently, for each remaining seru formation, we develop a branch-and-bound algorithm to obtain an optimal solution. In the branch-and-bound method, the lower bound is used not only to prune branches but also to eliminate the remaining seru formations. For large-scale instances, we propose a cooperative coevolution algorithm combined with simulated annealing that can generate nearly optimal solutions for large-scale problems in seconds. Extensive experiments demonstrate the superiority of the proposed algorithms, and tardy batches are reduced by 62.06% on average. Based on the experimental results, several managerial insights are provided to reduce tardy batches.

Application and Analysis of Hypergraph Association Rule Redundancy Algorithm in Data Mining

In order to realize “from individual data research to data system research” and “from passive data verification to active discovery,” this study proposes a hypergraph-based association rule redundancy processing algorithm in data mining. This study introduces the concepts of hypergraph and system, explores the establishment of hypergraph on a three-dimensional matrix model, and adopts a new hyperedge definition method according to the characteristics of big data and the concept of the system, which improves the ability to deal with problems; the association rules are transformed into a directed hypergraph, and the adjacency matrix is redefined. The detection of redundancy and loops is transformed into the processing of connected blocks and cycles in the hypergraph. The experimental results show that two UCI datasets were selected, namely, the balloons dataset and the shuttle landing control dataset, in which the minimum support and minimum confidence of the balloons dataset are both 5%. The dataset has 4 attributes, and 18 association rules are obtained through the Aprior algorithm. Although the running time of the coevolution algorithm is slightly longer than that of the other two global optimization algorithms, the running time is completely within the acceptable range. Moreover, due to the effective introduction of the idea of coevolution, compared with the use of the other two algorithms for association rule mining, it not only has a better mining quality but also has a significant advantage in the ability to jump out of the local optimal solution, realizing the search of high-quality association rules in high-dimensional datasets. Conclusion. This model provides a new idea and method for the redundant processing of association rules.

A two-stage cooperative evolutionary algorithm for energy-efficient distributed group blocking flow shop with setup carryover in precast systems

As the main link of the rapidly developing prefabricated construction industry, the production of precast components (PCs) has become a research hotspot. Therefore, the distributed group flow shop scheduling problem with blocking and carryover sequence-dependent setup time constraints (DPGFSP-BCT) in precast systems is considered. To address this problem, first, a mixed integer linear model is presented. Second, a two-stage cooperative coevolutionary algorithm (TS-CCEA) is proposed to minimize both the makespan and total energy consumption (TEC). In TS-CCEA, two acceleration rules are designed to reduce computational efforts. Third, to diversify the population, different initialization methods are established for different populations. Based on the problem-specific knowledge of solution classification, the individuals of the group population and job population execute two neighborhood search algorithms. Subsequently, considering the TEC, a critical path based speed mutation strategy is proposed to further improve the exploitation ability. Furthermore, a reinitialization heuristic is developed to avoid premature convergence. Last, the performance of the TS-CCEA is verified after calibrating the parameters. The experimental results demonstrate the stability and effectiveness of the algorithm.

Study on the necessity and strategy of popularization of "near-zero emission" technology for thermal power generation.

With the increasingly severe environmental problems, cleaner production has become one of the necessary means to alleviate the problem. The "near-zero emission" technology of thermal power generation is an active attempt of traditional power generation enterprises to further improve the level of clean coal utilization and promote the flue gas emission from conventional management to lean management. The analysis and evaluation of "near-zero emission" technology of thermal power generation in China are conducive to enhancing the scientific understanding of the overall situation of air pollution in China and thus further analyzing the current situation and future development path of thermal power emissions. In this paper, the intelligent prediction algorithm is used to predict and evaluate the emission scale of soot, sulfur dioxide, nitrogen oxides, and other air pollutants in China and the thermal power industry in the future, which reflects the urgency of the promotion of "near-zero emission" of thermal power from the macro level. In addition, based on Evolutionary Game Theory (EGT) and coevolutionary algorithm, the strategy selection of several virtual power generation groups under different policy conditions is simulated around the promotion of "near-zero emission" of thermal power. The results show that the policy subsidy plays an important role in promoting technology popularization.

A model and cooperative co-evolution algorithm for identifying driver pathways based on the integrated data and PPI network

Driver pathways have been acknowledged to play critical roles in the initiation and progression of cancers►hence it is essential for precision medicine related studies to develop accurate and efficient methods to identify them. Although previous approaches have shown promising results by integrating multi-omics data, their preset artificial parameters may decrease the convenience to use and limit the application scalability. In this paper, a novel integration approach is proposed to incorporate four omics data, i.e., construct a weighted non-binary mutation matrix without presetting artificial parameters. A parameter-free identification model is put forward based on it. It takes advantage of the association between genes in the PPI network as well as balances the contribution of coverage and mutual exclusivity with the harmonic mean. Furthermore, a cooperative co-evolution algorithm is proposed for solving this model. In the algorithm, a particle swarm optimization algorithm suitable for solving combinatorial problems is presented. Three cooperative operators are devised to construct the cooperation among the populations and the swarm to increase the population diversity. Both real biological datasets and simulated ones were exerted to perform experimental comparisons among the proposed method and six other state-of-the-art ones. The gene sets identified by the presented method generally contain more genes involved in known signaling pathways than those obtained by the other methods. Simultaneously, both high accuracy and high efficiency of the proposed method were verified in experiments, making it practical in realistic applications and an effective supplementary tool to identify driver pathways.

A nonlinear model and an algorithm for identifying cancer driver pathways

The identification of driver pathways has attracted considerable attention in recent years due to the high significance of driver pathways in cancer related studies. In most identification methods, the linear measurement is adopted for evaluating exclusivity, and some preset parameters are introduced to balance the contributions of different measurements. However, both the linear measurement and the preset parameters may exert negative effects on the applications of these methods. In this study, a nonlinear function is devised to measure exclusivity, and a parameter free nonlinear maximum weight submatrix (NMWS) identification model is proposed by considering coverage, exclusivity and connectivity. A competitive co-evolution algorithm (CCA) is also put forward for solving the presented NMWS model. Experimental results on both real biological data and simulated one were used compare the identification performance of the presented method with that of seven state of the art ones. The pathway detected by the presented method not only contains more genes enriched in a known signaling pathway, but also has a stronger connectivity in Protein–Protein Interaction (PPI) network. Simultaneously, the high efficiency of the presented method makes it practical in realistic applications. All of which have been confirmed through a large number of experiments.

An equilibrium model of the electricity market considering the participation of virtual power plants

Led by the carbon peak and neutrality targets, there is a consensus for renewable energy to compete in the electricity market alongside conventional energy sources. Based on co-evolutionary algorithms, this paper proposes an equilibrium model of electricity market with the participation of virtual power plants composed of wind farms and distributed energy storage. The model constructs a secondary clearing mechanism with the aim of promoting hydropower consumption and simulates the game behavior of various players in a market environment from the perspective of power producers. A co-evolutionary algorithm is used to solve the market equilibrium point for the game model, and the optimal fitness function is used to judge whether the generator’s strategy combination is optimal. The simulation results validate the effectiveness of the market model and the co-evolutionary algorithm. The results show that the secondary clearance mechanism is effective in increasing the rate of wind power consumption and improving the overall welfare of society.

Bi-level optimization based two-stage market clearing model considering guaranteed accommodation of renewable energy generation

The existing electricity market mechanisms designed to promote the consumption of renewable energy generation complicate network participation in market transactions owing to an unfair market competition environment, where the low cost renewable energy generation is not reflected in the high bidding price of high cost conventional energy generation. This study addresses this issue by proposing a bi-level optimization based two-stage market clearing model that considers the bidding strategies of market players, and guarantees the accommodation of renewable energy generation. The first stage implements a dual-market clearing mechanism that includes a unified market for trading the power generations of both renewable energy and conventional energy units, and a subsidy market reserved exclusively for conventional generation units. A re-adjustment clearing mechanism is then proposed in the second stage to accommodate the power generation of remaining renewable energy units after first stage energy allocations. Each stage of the proposed model is further described as a bi-level market equilibrium problem and is solved using a co-evolutionary algorithm. Finally, numerical results involving an improved IEEE 39-bus system demonstrate that the proposed two-stage model meets the basic requirements of incentive compatibility and individual rationality. It can facilitate the rational allocation of resources, promote the economical operation of electric power grids, and enhance social welfare.

Optimal priority assignment for real-time systems: a coevolution-based approach

In real-time systems, priorities assigned to real-time tasks determine the order of task executions, by relying on an underlying task scheduling policy. Assigning optimal priority values to tasks is critical to allow the tasks to complete their executions while maximizing safety margins from their specified deadlines. This enables real-time systems to tolerate unexpected overheads in task executions and still meet their deadlines. In practice, priority assignments result from an interactive process between the development and testing teams. In this article, we propose an automated method that aims to identify the best possible priority assignments in real-time systems, accounting for multiple objectives regarding safety margins and engineering constraints. Our approach is based on a multi-objective, competitive coevolutionary algorithm mimicking the interactive priority assignment process between the development and testing teams. We evaluate our approach by applying it to six industrial systems from different domains and several synthetic systems. The results indicate that our approach significantly outperforms both our baselines, i.e., random search and sequential search, and solutions defined by practitioners. Our approach scales to complex industrial systems as an offline analysis method that attempts to find near-optimal solutions within acceptable time, i.e., less than 16 hours.

Design of a Novel Wireless NoC Architecture for Chip Multiprocessor

The Network on chip is a new communication method of System on Chip. It is the main part of multi-core technology. A large number of embedded cores can be combined on a single chip. Due to the multi-hop characteristics of communication, there are some problems in the design of traditional NoC, such as high latency and power consumption. Thus, to handle the communication problem of long-distance transmission of traditional NoC, wireless Networks on Chip (WiNoC) has been proposed. Using WiNoC to design the multi-core System on Chip can significantly reduce the latency and energy consumption of the network. The advantage of using wireless network on chip is to replace the multi-hop path of traditional network on chip with wireless single hop link. However, due to the limited space on the chip and the cost, it is very important to determine the optimal number and correct location of wireless hubs. In this paper, we propose a new method to obtain the optimum configuration of a WiNoC by using the Co-evolutionary Algorithms (CA) and a wireless NoC architecture based on mesh. This work examines the effect of VC and sub-net size on WiNoC performance efficiency and energy under different packet injection rate (PIR) and the number of cores (64, 144 and 256). The simulation experiment of the algorithm proposed in the whole architecture has been implemented on the Noxim platform.

Mixed-model assembly line balancing problem considering preventive maintenance scenarios: MILP model and cooperative co-evolutionary algorithm

Considering preventive maintenance (PM) scenarios in assembly line balancing (ALB) problem has been proven to be an effective way to promote line efficiency. However, previous research only focused on the single-model assembly line and ignored the mixed-model assembly line, resulting in the line cannot meet the growing customization needs of customers. This paper addresses the mixed-model ALB considering PM scenarios (MALB_PMs). A mixed-integer mathematical formulation is proposed to optimize the cycle time and task alteration. The cooperative co-evolutionary algorithm is introduced to simplify the large-scaled cases via the powerful divide-and-conquer architecture and is enhanced with four improvements. An archive is generated to save complete solutions with better performance and evaluate the fitness of solutions. A mixed-model variable step-size decoding method is designed to speed up the decoding process of the algorithm. One inter-population crossover operator is designed to enhance the communication among sub-problems. Four objective-oriented neighbor search operators are proposed to promote the convergence performance of the algorithm. Finally, three small-scale instances are employed to illustrate the application of the mathematical model, and 39 cases are designed to test the performance of the developed algorithm. Experiment results demonstrate that the proposed mathematical formulation can obtain the Pareto solutions of small-scaled instances; The developed algorithm outperforms other algorithms on three criteria; The Pareto front obtained by this algorithm is closer to the True Pareto front.

Android malware detection as a Bi-level problem

Malware detection is still a very challenging topic in the cybersecurity field. This is mainly due to the use of obfuscation techniques. To solve this issue, researchers proposed to extract frequent API (Application Programming Interface) call sequences and then use them as behavior indicators. Several methods aiming at generating malware detection rules have been proposed with the goal to come up with a set of rules that is able to accurately detect malicious code patterns. However, the rules generation process heavily depends on the training database content which will affect the detection rate of the model when confronted to new variants of malicious patterns. In order to assess a rule’s detection accuracy, we need to execute the rule on the whole malware database which makes the detection rule quality evaluation very sensitive to the database content. To solve this issue, we suggest in this paper to consider the detection rules generation process as a BLOP (Bi-Level Optimization Problem), where a lower-level optimization task is embedded within the upper-level one. The goal of the upper-level is to generate a set of detection rules in the form of: trees of combined patterns. Those rules are able to detect not only the real patterns from the base of examples but also the artificial patterns generated by the lower-level. The lower-level aims to generate a set of artificial malicious patterns that escape the rules of the upper-level. An efficient co-evolutionary algorithm is adopted as a search engine to ensure optimization at both levels. Such an automated competition between the two levels makes our new method BMD (Bi-level Malware Detection) able to produce effective detection rules that are capable of detecting new predictable malicious behaviors in addition to existing ones. Based on the statistical analysis of the experimental results, our BMD method has shown its merits when compared to several relevant state-of-the-art malware detection techniques on different Android malware datasets.

Co-Evolutionary Fuzzy Deep Transfer Learning for Disaster Relief Demand Forecasting

Relief demand forecasting is vital to the success of disaster relief operations, but it is associated with challenges including insufficient training samples, incomplete and imprecise inputs, and inaccurate demands. This article presents a co-evolutionary fuzzy deep transfer learning (CoFDTL) method for relief demand forecasting where different types of disasters (e.g., earthquake, typhoon, and flood) are considered as different tasks. CoFDTL consists of three stages. First, a deep fuzzy learning model is used to learn latent representation of the shared inputs of all tasks. Second, a co-evolutionary algorithm is used to simultaneously learn task-specific features and the shared regressor. Third, the shared regressor is re-trained based on the best solutions obtained for different tasks in the second stage. Experiments demonstrate that CoFDTL exhibits significant performance improvements over the selected popular fuzzy learning, deep learning, and transfer learning models. This article also reports the application of CoFDTL to two real-world disasters in China, 2018. The proposed CoFDTL that integrates fuzzy deep learning, transfer learning, and co-evolutionary learning can be used for many other complex multi-task transfer learning problems with insufficient samples and uncertain information.

An Effective Cooperative Co-Evolutionary Algorithm for Distributed Flowshop Group Scheduling Problems.

This article addresses a novel scheduling problem, a distributed flowshop group scheduling problem, which has important applications in modern manufacturing systems. The problem considers how to arrange a variety of jobs subject to group constraints at a number of identical manufacturing cellulars, each one with a flowshop structure, with the objective of minimizing makespan. We explore the problem-specific knowledge and present a mixed-integer linear programming model, a counterintuitive paradox, and two suites of accelerations to save computational efforts. Due to the complexity of the problem, we consider a decomposition strategy and propose a cooperative co-evolutionary algorithm (CCEA) with a novel collaboration model and a reinitialization scheme. A comprehensive and thorough computational and statistical campaign is carried out. The results show that the proposed collaboration model and reinitialization scheme are very effective. The proposed CCEA outperforms a number of metaheuristics adapted from closely related scheduling problems in the literature by a significantly considerable margin.