Dunwei Gong

China University of Mining Technology, Suchow, Jiangsu Sheng, China

Are you Dunwei Gong?

Claim your profile

Publications (58)18.08 Total impact

  • Xiangjuan Yao, Dunwei Gong, Yali Gu
    [Show abstract] [Hide abstract]
    ABSTRACT: Research on complex networks is becoming a very hot topic in recent years, among which node matching problem is an important issue. The aim of node matching problem is to find out the corresponding relations between the individuals of associated networks. Traditional node matching problem of networks always hypothesize that a proportion of matching nodes are known. However, if the ratio of matched nodes is very small, the matching accuracy of the remaining nodes cannot be evaluated accurately. What is more, we may have not any matched nodes for reference at all. In view of this, this paper established the mathematic model of node matching problem based on the adjacency matrixes of networks, and presented an evolutionary algorithm to solve it. The experimental results show that the proposed method can achieve satisfactory matching precision in the absence of any matched nodes.
    11/2014;
  • [Show abstract] [Hide abstract]
    ABSTRACT: Interval multi-objective optimization problems (IMOPs), whose parameters are intervals, are considerably ubiquitous in real-world applications. Previous evolutionary algorithms (EAs) aim at finding the well-converged and evenly-distributed Pareto front. An EA incorporating with a decision-maker (DM)׳s preferences was presented in this study to obtain a Pareto-optimal subset that meets the DM׳s preferences. In this algorithm, the DM׳s preferences in terms of the relative importance of objectives were interactively input, and the corresponding preferred regions were then obtained. Based on these regions, solutions with the same rank were further distinguished to guide the search towards the DM׳s preferred region. The proposed method was empirically evaluated on four IMOPs and compared with other state-of-the-art methods. The experimental results demonstrated the simplicity and the effectiveness of the proposed method.
    Neurocomputing 08/2014; 137:241–251. · 2.01 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Many-objective optimization problems are common in real-world applications, few evolutionary optimization methods, however, are suitable for them up to date due to their difficulties. We proposed a reference points-based evolutionary algorithm (RPEA) to solve many-objective optimization problems in this study. In RPEA, a series of reference points with good performances in convergence and distribution are generated according to the current population to guide the evolution. Furthermore, superior individuals are selected based on the assessment of each individual by calculating the distances between the reference points and the individual in the objective space. The algorithm was applied to four benchmark optimization problems and compared with NSGA-II and HypE. The results experimentally demonstrate that the algorithm is strengthened in obtaining Pareto optimal set with high performances.
    07/2014;
  • [Show abstract] [Hide abstract]
    ABSTRACT: Interactive genetic algorithm (IGA), combining a user's intelligent evaluation with the traditional operators of genetic algorithms, are developed to optimize those problems with aesthetic indicators. The evaluation uncertainties and burden, however, greatly restrict the applications of IGA in complicated situations. Surrogate model approximating to the evaluation of the user has been generally applied to alleviate the evaluation burden of the user. The evaluation uncertainties, however, are not taken into account in existing research, therefore, a weighted multi-output gaussian process is here proposed to build the surrogate model by incorporating the uncertainty so as to enhance the performance of IGA. First, an IGA with interval fitness evaluation is adopted to depict the evaluation uncertainty, and the evaluation noise is defined based on the assignment. With the evaluation noise, the weight of each training sample is calculated and used to train a gaussian process which has two outputs to approximate the upper and lower values of the interval fitness, respectively. The trained gaussian process is treated as a fitness function and used to estimate the fitness of individuals generated in the subsequent evolutions. The proposed algorithm is applied to a benchmark function and a real-world fashion design to experimentally demonstrate its strength in searching.
    Computational Intelligence in Dynamic and Uncertain Environments (CIDUE), 2013 IEEE Symposium on; 01/2013
  • Dunwei Gong, Gengxing Wang, Xiaoyan Sun
    [Show abstract] [Hide abstract]
    ABSTRACT: Many-objective optimization problems are very common and important in real-world applications, and there exist few methods suitable for them. Therefore, many-objective optimization problems are focused on in this study, and a set-based genetic algorithm is presented to effectively solve them. First, each objective of the original optimization problem is transformed into a desirability function according to the preferred region defined by the decision-maker. Thereafter, the transformed problem is further converted to a bi-objective optimization one by taking hyper-volume and the decision-maker's satisfaction as the new objectives, and a set of solutions of the original optimization problem as the new decision variable. To tackle the converted bi-objective optimization problem by using genetic algorithms, the crossover operator inside a set is designed based on the simplex method by using solutions of the original optimization problem, and the crossover operator between sets is developed by using the entropy of sets. In addition, the mutation operator of a set is presented to obey the Gaussian distribution and change along with the decision-maker's preferences. The proposed method is applied to five benchmark many-objective optimization problems, and compared with other six methods. The experimental results empirically demonstrate its effectiveness.
    Computational Intelligence (UKCI), 2013 13th UK Workshop on; 01/2013
  • Dunwei Gong, Jian Chen, Xiaoyan Sun
    [Show abstract] [Hide abstract]
    ABSTRACT: This paper proposes a novel interactive genetic algorithm. In this algorithm, the weight of each gene meaning unit is calculated based on the emergence frequency of the corresponding allele. By comparing the weights, we determine the similarity of two individuals. Individuals in the current generation are selected for evaluation by the user according to the similarities between them and the most preferred one in the former generation. The fitness of unevaluated individuals is estimated based on the information of all evaluated individuals in the current generation. The proposed algorithm is applied to a curtain evolutionary design system, and compared with existing typical methods. The experimental results validate that the proposed algorithm has advantages in reducing the user’s fatigue and improving the efficiency in exploration.
    Kongzhi Lilun Yu Yinyong/Control Theory and Applications 01/2013; 30(5).
  • Dunwei Gong, Xinfang Ji
    [Show abstract] [Hide abstract]
    ABSTRACT: Multi-objective optimization problems with interval parameters are ubiquitous and important, yet not many effective methods are available for solving them. To solve these problems, we propose a set-based evolutionary algorithm incorporated with decision-maker (DM)’s preferences to obtain a Pareto solution set which satisfies DM’s preferences. In this algorithm, the original optimization problem is first transformed into a tri-objective deterministic optimization problem with three performance indicators: hyper-volume, uncertainty and DM satisfaction. To solve the transformed problem, we employ a set-based Pareto dominance relation to compare different individuals. Individuals with the same rank are distinguished by using a specially designed extension measure incorporating DM’s preferences. Additionally, a set-based mutation and recombination scheme is suggested to generate an offspring with high performance. Four benchmark multiobjective optimization problems and a car cab design problem are used to evaluate the proposed method. Results are compared with those from other three methods. Conclusions indicate that the proposed method can obtain a Pareto solution set with a desirable compromise among the convergence, extension, uncertainty and the DM’s satisfaction.
    Kongzhi Lilun Yu Yinyong/Control Theory and Applications 01/2013; 30(11).
  • Jing Sun, Dunwei Gong, Xinfang Ji
    [Show abstract] [Hide abstract]
    ABSTRACT: Interval multi-objective optimization problems are ubiquitous and important in real-world applications. An interactive evolutionary algorithm incorporating an optimization-cum-decision-making procedure is presented to obtain the most preferred solution that fits a decision-maker (DM)’s preferences. In this algorithm, a preference direction is elicited by requesting the DM to select the worst one from a part of non-dominated solutions. A metric based on the above direction, which reflects the approximation performance of a candidate solution, is designed to rank different solutions with the same rank and preference. The proposed method is applied to four interval bi-objective optimization problems, and compared with PPIMOEA as well as an a posteriori method. The experimental results show the effectiveness and high efficiency of the proposed method.
    Kongzhi yu Juece/Control and Decision 01/2013; 28(4).
  • [Show abstract] [Hide abstract]
    ABSTRACT: An interactive genetic algorithm with evaluating individuals using variational granularity was presented in this study to effectively alleviate user fatigue. In this algorithm, multiple language sets with different evaluation granularities are provided. The diversity of a population described with the entropy of its gene meaning units is utilized to first choose parts of appropriate language sets to participate in evaluating the population. A specific language set for evaluating an individual is further selected from these sets according to the distance between the individual and the current preferred one. The proposed algorithm was applied to a curtain evolutionary design system and compared with previous typical ones. The empirical results demonstrate the strengths of the proposed algorithm in both alleviating user fatigue and improving the efficiency in search.
    Proceedings of the 19th international conference on Neural Information Processing - Volume Part III; 11/2012
  • Dunwei Gong, Tian Tian, Xiangjuan Yao
    [Show abstract] [Hide abstract]
    ABSTRACT: Generating test data covering multiple paths using multi-population parallel genetic algorithms is a considerable important method. The premise on which the method above is efficient is appropriately grouping target paths. Effective methods of grouping target paths, however, have been absent up to date. The problem of grouping target paths for generation of test data covering multiple paths is investigated, and a novel method of grouping target paths is presented. In this method, target paths are divided into several groups according to calculation resources available and similarities among target paths, making a small difference in the number of target paths belonging to different groups, and a great similarity among target paths in the same group. After grouping these target paths, a mathematical model is built for parallel generation of test data covering multiple paths, and a multi-population genetic algorithm is adopted to solve the model above. The proposed method is applied to several benchmark or industrial programs, and compared with a previous method. The experimental results show that the proposed method can make full use of calculation resources on the premise of meeting the requirement of path coverage, improving the efficiency of generating test data.
    Journal of Systems and Software 11/2012; 85(11):2531–2540. · 1.14 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Surrogate-assisted interactive genetic algorithms (IGAs) are found to be very effective in reducing human fatigue. Different from models used in most surrogate-assisted evolutionary algorithms, surrogates in IGA must be able to handle the inherent uncertainties in fitness assignment by human users, where, e.g., interval-based fitness values are assigned to individuals. This poses another challenge to using surrogates for fitness approximation in evolutionary optimization, in addition to the lack of training data. In this paper, a new surrogate-assisted IGA has been proposed, where the uncertainty in subjective fitness evaluations is exploited both in training the surrogates and in managing surrogates. To enhance the approximation accuracy of the surrogates, an improved cotraining algorithm for semisupervised learning has been suggested, where the uncertainty in interval-based fitness values is taken into account in training and weighting the two cotrained models. Moreover, uncertainty in the interval-based fitness values is also considered in model management so that not only the best individuals but also the most uncertain individuals will be chosen to be re-evaluated by the human user. The effectiveness of the proposed algorithm is verified on two test problems as well as in fashion design, a typical application of IGA. Our results indicate that the new surrogate-assisted IGA can effectively alleviate user fatigue and is more likely to find acceptable solutions in solving complex design problems.
    IEEE transactions on systems, man, and cybernetics. Part B, Cybernetics: a publication of the IEEE Systems, Man, and Cybernetics Society 09/2012; · 3.01 Impact Factor
  • Xiaoyan Sun, Dunwei Gong, Wei Zhang
    [Show abstract] [Hide abstract]
    ABSTRACT: Interactive genetic algorithms are effective methods of solving optimization problems with implicit (qualitative) criteria by incorporating a user's intelligent evaluation into traditional evolution mechanisms. The heavy evaluation burden of the user, however, is crucial and limits their applications in complex optimization problems. We focus on reducing the evaluation burden by presenting a semi-supervised learning assisted interactive genetic algorithm with large population. In this algorithm, a population with many individuals is adopted to efficiently explore the search space. A surrogate model built with an improved semi-supervised learning method is employed to evaluate a part of individuals instead of the user to alleviate his/her burden in evaluation. Incorporated with the principles of the improved semi-supervised learning, the opportunities of applying and updating the surrogate model are determined by its confidence degree in estimation, and the informative individuals reevaluated by the user are selected according to the concept of learning from mistakes. We quantitatively analyze the performance of the proposed algorithm and apply it to the design of sunglasses lenses, a representative optimization problem with one qualitative criterion. The empirical results demonstrate the strength of our algorithm in searching for satisfactory solutions and easing the evaluation burden of the user.
    Applied Soft Computing. 09/2012; 12(9):3004–3013.
  • [Show abstract] [Hide abstract]
    ABSTRACT: Traditional methods of generating test data may result in redundancy of test data, which brings many troubles to software testing. In order to solve the redundancy of test data, this study proposed a novel approach of generating test data by reducing target statements based on dominant relations. First, basic concepts and principles concerning dominance are listed. Then, an approach is proposed to reduce target statements according to their dominant relations. Finally, test suite covering the reduced set of target statements is generated by a genetic algorithm. The generated test suite can also cover all original target statements, which is guaranteed by the proposed strategy. We applied the method to nine benchmark programs, and compared with traditional and greedy methods. The experimental results show that our method can not only reduce redundancy, but also improve the efficiency of generating test data.
    Evolutionary Computation (CEC), 2012 IEEE Congress on; 01/2012
  • Xiaoyan Sun, Lei Yang, Dunwei Gong, Ming Li
    [Show abstract] [Hide abstract]
    ABSTRACT: Interactive genetic algorithms (IGAs) have been successfully applied to optimize problems with aesthetic criteria by embedding the intelligent evaluations of a user into the evolutionary process. User fatigue caused by frequent interactions, however, often greatly impairs the potentials of IGAs on solving complicated optimization problems. Taking the benefits of collective intelligence into account, we here present an IGA with collective intelligence which is derived from a mechanism of group decision making. An IGA with interval individual fitness is focused here and it can be separately conducted by multiple users at the same time. The collective intelligence of all participated users, represented with social and individual knowledge, is first collected by using a modified group decision making method. Then the strategy of applying the collective intelligence to initialize and guide the single evolution of the IGA is given. With such a multi-user promoted IGA framework, the performance of a single IGA is expected to be evidently improved. In a local network environment, the algorithm is applied to a fashion design system and the results empirically demonstrate that the algorithm can not only alleviate user fatigue but also increase the opportunities of IGAs on finding most satisfactory solutions.
    Evolutionary Computation (CEC), 2012 IEEE Congress on; 01/2012
  • Dunwei Gong, Lei Yang, Xiaoyan Sun, Ming Li
    [Show abstract] [Hide abstract]
    ABSTRACT: Interactive genetic algorithms (IGAs) are effective methods of solving optimization problems with qualitative indices. The problem of user fatigue resulting from his/her evaluations, however, restricts their applications in complex optimization problems. Employing various surrogate models to evaluate (a part of) individuals instead of a user is a feasible approach to solving the above problem. Previous studies, however, have not fully utilized knowledge provided by users with similar preference when constructing these models. The problem of constructing surrogate models by using knowledge of users with similar preference was focused in this study. First, users with similar preference participating the evolution were identified based on the matrix formed by the relationship between users and the “fitness” of allele meaning units and the users' interests in allele meaning units by using the collaborative filtering algorithm based on nearest-neighbor; and then the individuals evaluated by users with similar preference and chosen according to the users' preference similarities and confidence, along with their fitness, were as a part of samples for training the surrogate model of the current user's cognition. The proposed method was applied to an evolutionary fashion design system, and the experimental results show that the proposed method can improve the capability in exploration on the premise of greatly alleviating user fatigue.
    Evolutionary Computation (CEC), 2012 IEEE Congress on; 01/2012
  • Lei Yang, Dunwei Gong, Xiaoyan Sun, Jing Sun
    [Show abstract] [Hide abstract]
    ABSTRACT: Interactive genetic algorithms (IGAs) are effective methods of tackling optimization problems involving qualitative indices by incorporating a user's evaluations into traditional genetic algorithms. The problem of user fatigue resulting from the user's evaluations, however, has a negative influence on the performance of these algorithms. Substituting the user's evaluations with various surrogate models is beneficial to alleviate user fatigue. Previous studies, however, have not taken full advantage of information provided by samples obtained earlier when constructing or updating these models. We focus on the issue of user fatigue in this study, and present a novel method of effectively alleviating user fatigue by substituting the user's evaluations with a weighted support vector machine (WSVM) and by incorporating it with the mechanism of transfer learning. The proposed method is applied to the fashion evolutionary design system and compared with previous effective IGAs. The experimental results confirm the advantage of the proposed method in both alleviating user fatigue and improving the precision of the surrogate model.
    Intelligent Control and Automation (WCICA), 2012 10th World Congress on; 01/2012
  • Source
    Dunwei Gong, Xiangjuan Yao
    [Show abstract] [Hide abstract]
    ABSTRACT: The application of genetic algorithms in automatically generating test data has become a research hotspot and produced many results in recent years. However, its applicability is limited in the presence of flag variables. This issue, known as the flag problem, has been studied by many researchers to date. We propose a novel method of testability transformation to tackle the flag problem. Different from traditional transformation methods, in our method, the key step is not the transformation of source code, but that of target statements. We search for a new target statement (or a set of target statements) equivalent to the original one and then transform the problem of generating test data that cover the original target statement into the one that cover the new target statement (or a set of target statements). We apply our method in many real-world programs, and the experimental results show its effectiveness. KeywordsSoftware testing–Flag problem–Target statement–Testability transformation
    Neural Computing and Applications 11/2011; · 1.76 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Problems with hybrid indices are common, but the performance of previous methods in solving these problems needs to be further improved. We propose an adaptive evolutionary optimization algorithm of solving the above problems effectively in this study. First, the convergence rate of a population is calculated based on the distance between the optimal individuals with adjacent generations; then the crossover and mutation probabilities are adjusted dynamically according to the diversity, the convergence rate of a population and the number of generations. We apply the proposed algorithm to an interior layout design problem, a typical optimization problem with hybrid indices, and compare it with the algorithm with constant crossover and mutation probabilities. The experimental results confirm that the proposed algorithm has advantages in the number, quality and distribution of optimal solutions. The proposed algorithm has a good tradeoff between exploration and exploitation, and provides an efficient way to solve an optimization problem with hybrid indices.
    Nature and Biologically Inspired Computing (NaBIC), 2010 Second World Congress on; 01/2011
  • Source
    Dunwei Gong, Wanqiu Zhang, Xiangjuan Yao
    [Show abstract] [Hide abstract]
    ABSTRACT: Highlights • The first study to consider the problem of generating test data for many target paths. • The first study to establish the optimization model for the problem of generating test data for many paths coverage. • Propose a strategy of grouping target paths in order to reduce the difficulty of the problem. • Propose a strategy of reconstructing sub-optimization problems during the evolution to reduce the difficulty further.
    Journal of Systems and Software 01/2011; 84:2222-2233. · 1.14 Impact Factor
  • Jing Sun, Dunwei Gong, Xiaoyan Sun
    [Show abstract] [Hide abstract]
    ABSTRACT: Interval multi-objective optimization problems (MOPs) are popular and important in real-world applications. We present a novel interactive evolutionary algorithm (IEA) incorporating an optimization-cum-decision-making procedure to obtain the most preferred solution that fits a decision-maker (DM)'s preferences. Our method is applied to two interval MOPs and compared with PPIMOEA and the posteriori method, and the experimental results confirm the superiorities of our method.
    Neural Information Processing - 18th International Conference, ICONIP 2011, Shanghai, China, November 13-17, 2011, Proceedings, Part II; 01/2011