Research article

A survey comparative analysis of cartesian and complexity science frameworks adoption in financial risk management of Zimbabwean banks

  • Traditionally, financial risk management is examined with cartesian and interpretivist frameworks. However, the emergence of complexity science provides a different perspective. Using a structured questionnaire completed by 120 Risk Managers, this paper pioneers a comparative analysis of cartesian and complexity science theoretical frameworks adoption in sixteen Zimbabwean banks, in unique settings of a developing country. Data are analysed with descriptive statistics. The paper finds that overally banks in Zimbabwe are adopting cartesian and complexity science theories regardless of bank size, in the same direction and trajectory. However, adoption of cartesian modeling is more comprehensive and deeper than complexity science. Furthermore, due to information asymmetries, there is diverging modeling priorities between the regulator and supervisor. The regulator places strategic thrust on Knightian risks modeling whereas banks prioritise ontological, ambiguous and Knightian uncertainty measurement. Finally, it is found that complexity science and cartesianism intersect on market discipline. From these findings, it is concluded that complexity science provides an additional dimension to quantitative risk management, hence an integration of these two perspectives is beneficial. This paper makes three contributions to knowledge. First, it adds valuable insights to theoretical perspectives on Quantitative Risk Management. Second, it provides empirical evidence on adoption of two theories from developing country perspective. Third, it offers recommendations to improve Quantitative Risk Management policy formulation and practice.

    Citation: Gilbert Tepetepe, Easton Simenti-Phiri, Danny Morton. A survey comparative analysis of cartesian and complexity science frameworks adoption in financial risk management of Zimbabwean banks[J]. Quantitative Finance and Economics, 2022, 6(2): 359-384. doi: 10.3934/QFE.2022016

    Related Papers:

    [1] Yutao Lai, Hongyan Chen, Fangqing Gu . A multitask optimization algorithm based on elite individual transfer. Mathematical Biosciences and Engineering, 2023, 20(5): 8261-8278. doi: 10.3934/mbe.2023360
    [2] Shijing Ma, Yunhe Wang, Shouwei Zhang . Modified chemical reaction optimization and its application in engineering problems. Mathematical Biosciences and Engineering, 2021, 18(6): 7143-7160. doi: 10.3934/mbe.2021354
    [3] Shihao Yuan, Hong Zhao, Jing Liu, Binjie Song . Self-organizing map based differential evolution with dynamic selection strategy for multimodal optimization problems. Mathematical Biosciences and Engineering, 2022, 19(6): 5968-5997. doi: 10.3934/mbe.2022279
    [4] Xiaofang Guo, Yuping Wang, Haonan Zhang . An active learning Gaussian modeling based multi-objective evolutionary algorithm using population guided weight vector evolution strategy. Mathematical Biosciences and Engineering, 2023, 20(11): 19839-19857. doi: 10.3934/mbe.2023878
    [5] Jiande Zhang, Chenrong Huang, Ying Huo, Zhan Shi, Tinghuai Ma . Multi-population cooperative evolution-based image segmentation algorithm for complex helical surface image. Mathematical Biosciences and Engineering, 2020, 17(6): 7544-7561. doi: 10.3934/mbe.2020385
    [6] Wenqiang Zhang, Xiaoxiao Zhang, Xinchang Hao, Mitsuo Gen, Guohui Zhang, Weidong Yang . Multi-stage hybrid evolutionary algorithm for multiobjective distributed fuzzy flow-shop scheduling problem. Mathematical Biosciences and Engineering, 2023, 20(3): 4838-4864. doi: 10.3934/mbe.2023224
    [7] Wenqiang Zhang, Chen Li, Mitsuo Gen, Weidong Yang, Zhongwei Zhang, Guohui Zhang . Multiobjective particle swarm optimization with direction search and differential evolution for distributed flow-shop scheduling problem. Mathematical Biosciences and Engineering, 2022, 19(9): 8833-8865. doi: 10.3934/mbe.2022410
    [8] Jiahao Zhang, Zhengming Gao, Suruo Li, Juan Zhao, Wenguang Song . Improved intelligent clonal optimizer based on adaptive parameter strategy. Mathematical Biosciences and Engineering, 2022, 19(10): 10275-10315. doi: 10.3934/mbe.2022481
    [9] Margaritis Kostoglou, Thodoris Karapantsios, Maria Petala, Emmanuel Roilides, Chrysostomos I. Dovas, Anna Papa, Simeon Metallidis, Efstratios Stylianidis, Theodoros Lytras, Dimitrios Paraskevis, Anastasia Koutsolioutsou-Benaki, Georgios Panagiotakopoulos, Sotirios Tsiodras, Nikolaos Papaioannou . The COVID-19 pandemic as inspiration to reconsider epidemic models: A novel approach to spatially homogeneous epidemic spread modeling. Mathematical Biosciences and Engineering, 2022, 19(10): 9853-9886. doi: 10.3934/mbe.2022459
    [10] Hong Lu, Hongxiang Zhan, Tinghua Wang . A multi-strategy improved snake optimizer and its application to SVM parameter selection. Mathematical Biosciences and Engineering, 2024, 21(10): 7297-7336. doi: 10.3934/mbe.2024322
  • Traditionally, financial risk management is examined with cartesian and interpretivist frameworks. However, the emergence of complexity science provides a different perspective. Using a structured questionnaire completed by 120 Risk Managers, this paper pioneers a comparative analysis of cartesian and complexity science theoretical frameworks adoption in sixteen Zimbabwean banks, in unique settings of a developing country. Data are analysed with descriptive statistics. The paper finds that overally banks in Zimbabwe are adopting cartesian and complexity science theories regardless of bank size, in the same direction and trajectory. However, adoption of cartesian modeling is more comprehensive and deeper than complexity science. Furthermore, due to information asymmetries, there is diverging modeling priorities between the regulator and supervisor. The regulator places strategic thrust on Knightian risks modeling whereas banks prioritise ontological, ambiguous and Knightian uncertainty measurement. Finally, it is found that complexity science and cartesianism intersect on market discipline. From these findings, it is concluded that complexity science provides an additional dimension to quantitative risk management, hence an integration of these two perspectives is beneficial. This paper makes three contributions to knowledge. First, it adds valuable insights to theoretical perspectives on Quantitative Risk Management. Second, it provides empirical evidence on adoption of two theories from developing country perspective. Third, it offers recommendations to improve Quantitative Risk Management policy formulation and practice.



    Evolutionary multitasking optimization is inspired by the extraordinary talent of people to perform multiple tasks simultaneously [1]. In 2016, Gupta et al. [2] introduced the multifactor genetic mechanism into evolutionary algorithms and proposed the multifactor evolutionary algorithm (MFEA) for the first time, which is considered to be the pioneering work in multitasking evolutionary algorithms. Compared with traditional single-task evolutionary algorithms, MFEA shows superiority, but MFEA suffers from slow convergence, easy to trap in local optimum local optimality, and negative migration. Subsequently, researchers have conducted extensive studies on EMTO and have made some progress. Liang et al. [3] introduced hyper-rectangular search and gene mapping strategies into MFEA to construct similar gene representation space while expanding the search space to improve transfer efficiency and accelerate population convergence. Bali et al. [4] introduced a linear-domain adaptive strategy of search space to improve MFEA (LDA-MFEA), which reduces inter-task differences and facilitates positive migration between tasks. Ong et al. [5] introduced online parameter learning in MFEA to adaptively adjust the strength of transfer knowledge and reduce negative migration. Yang et al. [6] used K-means clustering to cluster individuals in a population and selected different random interaction probabilities for different categories of variables, improving the efficiency of knowledge transfer between tasks. Liaw et al. [7] considered the information exchange between all tasks as a symbiotic relationship and proposed a generalized framework, SBO, that is applicable to multitasking problems. Feng et al. [8] applied a two-population framework to perform mapping from source domain to target domain via a single-layer denoising autoencoder for explicit knowledge transfer among tasks. Li et al. [9] developed the multi-population multitasking evolutionary optimization framework (MPEF), which introduces a multi-population evolutionary framework into multitasking optimization, allowing different optimization tasks to be handled by different evolutionary algorithms. Cai et al. [10] adopted a hybrid knowledge transfer strategy to select the corresponding transfer strategy according to the correlation between tasks.

    In EMTO, the identification of valuable knowledge between tasks, that is, the selection of transfer content, is a factor affecting the performance of the EMTO algorithm. Existing research has shown that transferring elite solutions between two relevant tasks may accelerate the convergence of algorithms, but if these elites are local optima, they will lead the population to the local optima [11]. Therefore, determining the valuable content to transfer between tasks is an important issue. Recently, in order to improve the performance of multi-task optimization algorithms, many methods have been proposed to identify valuable knowledge between tasks. Gao et al. [12] proposed an EMTO algorithm based on semi-supervised learning that identifies individuals containing useful knowledge and selects them for transfer between tasks. In [13], the individuals in the source task that are superior to the optimal individuals in the target task are chosen to constitute the transferred population. Then, the excellent individuals in the transfer population are combined with the target population to construct a Gaussian mixture distribution model and generate the offspring according to the mixture model to realize the knowledge sharing between tasks. Lin et al. [14] used the neighbors of positive transferred individuals obtained during the evolutionary process as the next generation of transfer solutions, achieving automatic identification of valuable transfer solutions during the evolutionary process. Sun et al. [15] used solutions in the source task that were similar to non-dominated solutions in the target task as transfer solutions. In [16], the authors utilized an incremental naïve Bayes classifiers to determine the individuals to transfer. In [17], the authors used anomaly detection to determine the individuals to transfer, and identified individuals carrying valuable knowledge, transferring them to the target task. Chen et al. [18] introduced the concept of transfer rank to quantify the priority of individuals and select transferred content based on the quantified priority.

    Although these algorithms have shown the potential to solve the EMTO problem, their ability to find valuable transferred solutions needs to be further improved. The main purpose of EMTO is to transfer useful knowledge between tasks to facilitate the convergence of the target task, assuming that the co-optimized tasks are related. However, in reality many optimization problems are black-box optimizations without prior knowledge of task similarity. Therefore, how to identify and select knowledge to transfer between tasks becomes a key factor affecting the performance of EMTO. In addition, since evolutionary algorithms are based on populations, which contain rich information about data distribution, extracting and utilizing the feedback information of populations in the evolutionary process for the design of the EMTO algorithm helps to improve the performance of the algorithm. The elite solutions in the population are generally located near the global optimal solution; thus, finding solutions with similar distributions to the elites in the target task as the migration content in the resource task can accelerate the convergence of the target task.

    In order to achieve more effective and robust knowledge transfer between different optimization tasks, this article proposes a single-objective EMTO algorithm based on population distribution. The proposed algorithm uses differential algorithm as a task solver, namely, AMTDE-PD. In addition, an improved randomized mating probability is also included in the proposed algorithm to adjust the intensity of the knowledge transfer. The primary contributions of this article are twofold.

    1) The transfer content selection strategy based on population distribution information is proposed. This strategy can adaptively select the knowledge to be transferred between tasks, and the transfer solutions may be not only elite solutions but also other solutions.

    2) Design an adaptive interaction probability to adjust the interaction intensity between tasks. The method adjusts the interaction probability between tasks based on their evolutionary trends to mitigate negative transfer between tasks.

    The rest of this article is organized as follows. Section 2 gives the preliminaries about EMTO, and DE algorithm. The implementation details of AMTDE-PD are presented in Section 3. Section 4 provides and analyzes the experimental results. Section 5 gives the conclusion of this article.

    EMTO algorithms use evolutionary algorithms as task solvers to exploit potential synergies or complementarities between tasks to simultaneously solve multiple tasks with different search spaces [2]. That is to say, the evolutionary multitasking optimization algorithm returns the optimal solutions for multiple tasks through one search. Assuming that there are Θ minimization single-objective problems to be optimized simultaneously, Tθ denotes the θth task, Xθ is the search space corresponding to the task Tθ, and its objective function is Fθ:XθR, R is the real set, the mathematical representation of the multitasking optimization problem can be described as below:

    {x1,x2,xΘ}=argmin{F1(x1),,FΘ(xΘ)}s.t.xθXθ (1)

    where x1 is the best solution of Tθ. xθ is a feasible solution in the search space Xθ.

    During multitasking optimization process, since each optimization task has its own search space, the evolutionary multitasking algorithm has to map these tasks corresponding to different search spaces to a unified search space before solving these tasks and optimizing them using the corresponding evolutionary algorithm. Suppose the dimension of the search space for the θth task is denoted as Dθ, the dimension of the unified search space U is given as: DU=max{D1,D2,,DΘ}, where U[0,1]DU. The solution xθ in task Tθ are encoded into the uniform search space by

    udθ=(xdθLdθ)(HdθLdθ) (2)

    where udθ is the dth dimension of the individual xθ in U. Hdθ and Ldθ denote the upper and lower bound of dth dimension of Xθ. Conversely, decoding uθ into the original search space via

    xdθ=udθ×(HdθLdθ)+Ldθ (3)

    The differential evolution (DE) algorithm is one of the most efficient evolutionary algorithms for solving continuous optimization problems [19]. In this paper, we use DE as a task solver. DE evolves a population by performing a random initialization in the search space and then performing mutation, crossover, and selection operations. A mutation strategy is used in the DE algorithm to generate a mutation vector for each xi in the population [20]. An advanced mutation strategy in DE is as follows:

    DE/currenttopbest/1:vi=xi+F.(xpbestxi)+F.(xr1˜xr2) (4)

    where vi is a mutant vector, xpbest is randomly chosen from the top p% individuals in the population, p(0,1]. xr1 is a random solution in the population. ˜xr2 is randomly chosen from the union of the population and its external archive Arc. Arc is used to preserve the solutions that were eliminated during the evolutionary process. F is the scaling factor.

    After performing the mutation operation, DE uses the following binomial crossover to generate a new trial vector ui [21]:

    ui,d={vi,d, if rand <CRiord==jrandxi,d, otherwise  (5)

    where CR is the crossover factor applied to determine the number of offspring individuals obtaining variables from the mutant vector vi, CR ∈ [0, 1], updated with reference to the literature [10]. rand is a random number between 0 and 1. jrand [1, D], and D is the dimension of the search space.

    After the crossover operation is executed, the following selection operation is used for the minimization problem to generate the next generation of new population individuals [22]:

    xi={ui,iff(ui)<f(xi)xi,otherwise (6)

    where f(ui) and f(xi) denote the objective function value of the trial vector ui and individual xi, respectively. This operation ensures that the newly generated population is at least not worse than the previous generation's population.

    Due to its simplicity and effectiveness, and the fact that it has been widely used for multitasking solving [23,24,25], AMTDE-PD uses DE as a task solver.

    In the last few years, EMTO algorithms have been widely used for solving various problems [26], vehicle routing problems [27,28,29,30], traveling salesman problems [31], path planning problems [32], and hyperspectral image classification [33]. However, the common knowledge between different tasks has not been fully explored in traditional single-objective EMTO algorithms. In this article, we present a single-objective multitasking optimization algorithm based on population distribution to use the common knowledge between tasks. In this section, we first give the main framework of AMTDE-PD. Next, the transfer content selection strategy and inter-task knowledge transfer strategy are introduced. Then, an improved adaptive mating probability strategy is presented. Finally, the improved adaptive mating probability strategy is given.

    The flowchart of AMTDE-PD is illustrated in Figure 3. Algorithm 1 shows the main framework of AMTDE-PD. First, a randomly initialized population containing N individuals is assigned to each task, and all individuals in the population are evaluated on the corresponding task. At the beginning of each iteration, the subpopulation division operation is performed on target and resource tasks, respectively. Subsequently, the transferred population is constructed based on subpopulation (lines 6 and 7). Then, N offspring were generated for each task using two different methods (lines 8–17). At the end of each iteration, the total successful evolution rate SRt of the target population and the center-position distance between the target population and the resource population are calculated (lines 18 and 19). Next, the random interaction probability between tasks is updated based on the change in the distance of the center location between populations. To adjust the interaction probability RMPt of the current task Tt,(tΘ), we first check whether the total evolutionary success rate SRt (the number of offspring in the population that are better than their parents/the population size) of the population Pt is larger than a pre-specified threshold δ. If SRt>δ, it means that the population can achieve better evolution using the current RMP, and there is no need to update the RMP; otherwise, if SRt<δ, it means that the RMP between the populations needs to be adjusted. Finally, the optimal solutions for all tasks are output.

    Figure 1.  The flowchart of AMTDE-PD.
    Figure 2.  The process of calculating the MMD value between CKse and CKtB.
    Figure 3.  (a) Population distribution of tasks in the initial phase (b) Evolutionary trends of similar tasks and (c) Evolutionary trends of dissimilar tasks.

    Algorithm 1 Main Framework of AMTDE-PD
    Input: N: population size of each task;
         Θ: number of tasks;
         T1,TΘ: all tasks;
         K: number of clusters;
         MaxFES: maximum function evaluation;
         RMP0: initial value of interaction probability for each task;
         δ: The threshold for interaction probability adjustment;
         q: control parameters;
    Output:the optimal solutions for all tasks
    1 Initialize and evaluate each population Pθ(θ=1,2,,Θ);
    2 FES = ΘN; g = 1
    3 whileFESMaxFESdo
    4   Perform subpopulation division operations for target and resource tasks; (see Algorithm 2)
    5   The MMD values of the clusters where the optimal solution of the target task is located and all
    the clusters in the source task are calculated using Eq (10);
    6   Use Eqs (11) and (12) to construct the transferred population;
    7   fort=1:Θ
    8     fori=1:N
    9       ifrand<RMPt
    10         Use Eq (13) to generate the mutant individual vti;
    11       else
    12         Use Eq (14) to generate the mutant individual vti;
    13       endif
    14       Perform the crossover operation on xti and vti using Eq (5) to produce a trial vector ui and evaluate ui on task Tt;
    15       Perform selection operation using Eq (6) to generate the individual of the next generation population;
    16     endfor
    17     Calculate the total successful evolution rate SRt of the population;
    18     Calculate the center-position distance Distt,s between target task Tt and source task Tl according to Eq (7);
    19     ifSRt<δ
    20       RMPt(g+1)=Update(RMPt(g),Distt,s,q) (see Algorithm 3);
    21     endif
    22   end for
    23   FES=FES+N;
    24   g=g+1;
    25 endwhile

    Transferring the elite solution in a task as a transfer solution between tasks is effective if the co-optimized task pairs are correlated. However, there is no a priori knowledge of inter-task correlation. Therefore, this method of choosing what to transfer will result in a negative transfer if task pairs are irrelevant or of low relevance. An elite solution in one task may be inferior in another. Herein, a transfer content selection strategy based on population distribution information is proposed to select appropriate transferred content between tasks. First, each task population was divided into K sub-populations based on the fitness values of the individuals. Second, the Maximum Mean Discrepancy (MMD) is used to select the most similar clustering from the source task to the clustering of the optimal solution of the target task as the transferred knowledge between tasks. The steps are as follows:

    1) Division of subpopulations

    Individuals in each task population are first sorted in ascending order of fitness, and then individuals are categorized into advantaged, intermediate, and disadvantaged solutions a based on their fitness values. Then, the task population is categorized into three sub-populations: superior, intermediate, and inferior. The pseudocode for the subpopulation division process is presented in Algorithm 2.

    Algorithm 2 Subpopulation division process
    Input:Task populations P1,,PΘ; Number of subpopulation:K; Task population size:popt
    Output: K subpopulation
    1 Individuals in the population were sorted in ascending order of fitness values;
    2 Calculate grouping factors λ=popt/K;
    3 Calculate the size of each subpopulation based on the grouping factor;
    4 Task population is divided into K subpopulations according to the size of the subpopulation;

    2) Calculate the MMD value between each cluster in the source task and the cluster where the optimal solution of the target task lies

    MMD is a method to determine whether two data distributions are similar by measuring the distance between the two distributions in the reproducing kernel Hilbert space (RKHS) [34]. In [35], MMD is used to select one of the source domains that is most similar to the target domain for inter-task knowledge transfer and has achieved significant success. Inspired by this, we utilize MMD to compute the distance of the clusters and get the similarity between clusters based on the MMD value. When calculating the MMD value, the two high-dimensional distributions are mapped to the RKHS, then the distribution difference is calculated. A smaller MMD value shows that the two distributions are less different, that is, the search preferences of the two clusters are more similar. Suppose the dataset X=(x1,,xμ) and Y=(y1,,yv) obey the distributions P and Q, respectively. F is a class of functions f:χR that maps the original search space χ to the set R of real numbers. Let RKHS be denoted as H, φ:χH denotes a mapping from χ to H. The following is the inner product representation of f(x).

    f(x)=f,φ(x)H (8)

    The MMD values of P and Q in RKHS are calculated as follows:

    MMD(F,P,Q)=supfH1EP[f(x)]EQ[f(y)]=1μμi=1φ(xi)1ννj=1φ(yj)2 (9)

    At each generation, the MMD value between two cluster distributions is calculated as follows:

    MMD(CKse,CKtB)=1NNi=1φ(cksi)1MMj=1φ(cktj)2,e=1,2,3 (10)

    where CKse is a subpopulation in the source task population, CKtB is the subpopulation where the optimal solution in the target task lies cksiCKse, cktjCKtB. The process of calculating the MMD value is shown in Figure 2.

    As can be seen from Figure 2, the MMD values between CKtB and all subpopulations in the resource task are calculated, and the subpopulation in the resource task corresponding to the minimum MMD value is selected as the migration population.

    1) Calculate the index of the cluster CKse corresponding to the smallest MMD value

    idx=arg(min1e3(MMD(CKse,CKtB))) (11)

    2) Transferred content selection

    Individuals in the clusters indexed as idx in the source task are used as transferred content in the transfer population TP.

    TPCKsidx (12)

    Suppose there is a multitasking optimization problem containing Θ tasks. During the evolutionary process, if the Random Mating Probability (RMP) between tasks satisfies the condition rand<RMPθ(θ=1,2,.,Θ), then inter-task evolution is performed. Otherwise, intra-task self-evolution is executed. AMTDE-PD uses the DE algorithm as the task solver and designs the transfer strategy based on the mutation strategy in the DE algorithm. In this paper, we propose a mutation strategy DE/randtopbest that considers both exploration and exploitation. In AMTDE-PD, the transfer strategy is designed according to DE/randtopbest. Let Tt be the target task, the inter-task transfer strategy is designed based on the following:

    vti=xsr1+F.(xtpbestxsr1)+F.(xsr2xsr3) (13)

    where vti is the mutant individual corresponding to the current individual xti in the target task; xsr1, xsr2, and xsr3 are three mutually exclusive individuals randomly selected from the transfer population TP. xtpbest is the individual with the top 100p, p(0,1] fitness values in the target task. F is the scale factor and is updated with reference to [20].

    When the condition rand<RMPθ is not satisfied, AMTDE-PD performs intra-task self-evolution using the mutation strategy DE/currenttopbest/1 and binomial crossover, specifically as follows:

    vti=xti+F(xtpbest xti)+F(xtr1˜xtr2) (14)

    where xtr1 is randomly selected from the target task. ˜xtr2 is randomly selected from the union of the target population and its external archive.

    At each generation, AMTDE-PD first judges the condition rand<RMP is satisfied and then decides whether to perform inter-task evolution or intra-task self-evolution.

    In the EMTO algorithm, multiple tasks corresponding to different search spaces are mapped to a unified search space. In the early stages of evolution, solutions from different tasks are randomly distributed in a unified search space. As evolution progresses, related tasks move closer together while unrelated tasks move further apart. Figure 3 illustrates this situation.

    In the evolutionary multitasking optimization algorithm, RMP is used to control the interaction intensity between tasks. In existing studies, RMP is usually adjusted adaptively based on the improvement rate of fitness values during the evolution process. This method can promote the interaction between tasks with similar fitness landscapes and mitigate the impact of negative transfer between tasks with dissimilar fitness landscapes. However, when jointly optimizing two unrelated tasks, this approach can lead to negative transfer. As evolution progresses, the search domains of two unrelated tasks gradually overlap less and eventually become completely non-overlapping, as depicted in Figure 3(c). In this case, knowledge transfer between tasks becomes ineffective and consumes a large number of computational resources.

    In view of this, we adjust the RMP between tasks based on the evolutionary trend of populations during the evolution process. If two task populations evolve in the same direction, the intensity of knowledge transfer between tasks is enhanced to accelerate convergence. Otherwise, the intensity of knowledge transfer between tasks is weakened to avoid negative transfer. The direction of change in the population center position can reflect the evolutionary trend of the population, so the evolutionary trend between tasks is estimated by comparing distances between tasks' population centers in current and previous generations. Let the center positions of the target task and source task be denoted as Xt and Xs, respectively. In generation g, the distance of the center positions between the target and source tasks is calculated as:

    Distt,s(g)=XtXs2 (15)

    The calculation method for the center position of the population is as follows:

    X=1NNi=1xi (16)

    where N is the size of the population and xi is the individuals in the population.

    If Distt,s(g)>Distt,s(g1), it means that the populations of the two tasks are moving away from each other, the evolutionary tendency between the tasks is opposite, and decreasing RMP reduces the interaction strength between the tasks. If Distt,s(g)<Distt,s(g1), it means that the populations of the two tasks are approaching each other, and the evolutionary tendency between the tasks is the same, so increase the RMP to improve the interaction strength between the tasks. However, when RMP is greater than or equal to 1, the evolution between tasks will occupy all computational resources, and intra-task self-evolution will not be able to proceed, which is detrimental to the evolution of the population. Therefore, when [RMP]_t is greater than or equal to 1, it should be set back to 0.5 to ensure a balance between inter-task evolution and intra-task self-evolution. The pseudo-code of the update of RMP is shown in Algorithm 3.

    Algorithm 3 Update of RMP
    Input: Inter-task mating probability for generation RMPt(g); Distance between population center positions Distt,s; Control parameters q.
    Output:RMPt(g+1)
    1 if Distt,s(g)<Distt,s(g1)
    2   RMPt(g+1)=RMPt(g)/q;
    3   ifRMPt(g+1)1
    4     RMPt(g+1)=0.5;
    5   end if
    6   else
    7     RMPt(g+1)=RMPt(g)q;
    8 endif

    To evaluate the search efficiency of AMTDE-PD, we compare it with other advanced EMTO algorithms on two single-object multitasking test suites. The first test suite CEC2019-SOMTP [36] is the single-objective multitasking standard test set, which comprises nine single-objective MTO problems, each consisting of two minimization tasks. According to the intersection of the global optimum of the optimization task these problems can be classified into three categories: Complete Intersection (CI), Partial Intersection (PI), and No intersection (NI). In addition, Spearman's correlation coefficient (SRCC) was used to assess the similarity between the fitness landscapes of two tasks. According to the results of SRCC, these problems can be further characterized into three categories: High Similarity (HS), Medium Similarity (MS), and Low Similarity (LS). The second test suite is the single-objective multi-task complex test set WCCI2020-SOCTP1 contains ten complex single-objective MFO problems. Different from the CEC2019-SOMTP, the task functions composing the complex test set are obtained by rotating and shifting four hybrid functions and seven multi-mode functions in the CEC2014 test suite to varying degrees, which makes them more difficult to solve.

    1http://www.bdsc.site/websites/MTO/index.html

    To assess the effectiveness of the algorithm, we compare AMTDE-PD with nine advanced EMTO algorithms on the CEC2019-SOMTP, including four classical algorithms using EA as solver: MFEA [2], GMFEA [37], MTGA [38], SBO [7], and LDA-MFEA [4]; three high-performance algorithms using DE as solvers: MFDE [39], MFMP [9], and MPEFMTO [40]; and one algorithm that uses both DE and EA as task solvers: EMT-EAE [8]. Another reason for choosing these algorithms is that these algorithms use elite-based solutions, and randomized solutions, respectively, as transferred content. To further validate the superiority of the AMTDE-PD algorithm in solving complex optimization problems, the AMTDE-PD algorithm is compared with MFEA, MFDE, MFMP, and MPEFMTO on the CEC2019-SOMTP test suite.

    To ensure an unbiased comparison, the maximum function evaluation of each algorithm was set to 100,000, and each algorithm was run independently 20 times on each test problem. All other parameters in the comparison algorithms obey their original paper. The parameters of AMTDE-PD on the two test suites are set as follows: the population size of each task N = 100, the number of clusters K = 3, the RMP update threshold δ=0.5, the control parameter q=0.9, the initial values of both F and CR are set to 0.5.

    Synergy performance metrics proposed in the literature [38] are used in this experiment to verify the comprehensive performance of an algorithm on multiple tasks in a multitasking environment. Let the test case contain Θ minimization tasks. The optimal solution obtained by the lth execution of an algorithm on the task Tθ,θ(1,2,,Θ) is denoted as fir,θ. Let each algorithm be executed R times, then the performance metric of the ith algorithm is defined as follows:

    Scorei=Θθ=1Rr=1fiθ,rμθσθ (17)

    where μθ and σθ are the mean and variance of fir,θ obtained by all algorithms running R times on task Tθ, respectively.

    For the minimization problem, the smaller Score value of the algorithm indicates the better overall performance of the corresponding EMTO algorithm on the multitasking problem.

    The AMTDE-PD algorithm and the nine comparison algorithms are executed independently 20 times on the CEC2019-SOMTP, and the mean and standard deviation of the optimal function values achieved from the solution are shown in Table 1. The results of the Wilcoxon rank sum test at the significance level α = 0.05 are given in the last row of Table 1, and the symbols "+", "-" and "≈" indicate that the comparison algorithms' performance is better than, worse than, and approximate to the AMTDE-PD algorithm. To further minimize statistical comparison errors, Friedman's test is used to examine the difference between AMTDE-PD and the other comparison algorithms. The average rank is recorded as the comparison results, and a smaller one indicates better performance. The results of Friedman's test at the 0.05 significance level are given in Table 2, and the best results are shown in bold. Furthermore, this section uses the algorithm's performance metric Score as an assessment of the algorithm's performance, which reflects the algorithm's comprehensive performance on a problem. The values of the performance metric Score for the AMTDE-PD algorithm and the other compared algorithms are given in Table 3.

    Table 1.  Experimental results of AMTDE-PD algorithm and other nine comparison algorithms on CEC2019-SOMTP test suite.
    Problem Task MFEA MFDE GMFEA LDA-MFEA SBO MTGA EMT-EAE MPEF MPEFMTO AMTDE-PD
    CI+HS T1 3.74E-01-(6.64E-02) 8.94E-04-(2.70E-03) 1.05E+00-(1.59E-02) 3.38E-01-(7.60E-02) 9.35E-01-(5.64E-02) 1.24E-02-(8.83E-03) 7.05E-01-(8.22E-02) 1.93E-08-(1.55E-08) 2.58E-09-(2.54E-09) 4.7968e-12(7.64E-12)
    T2 1.98E+02-(5.16E+01) 1.89E+00-(5.65E+00) 3.68E+02-(2.57E+01) 1.53E+02-(3.19E+01) 3.09E+02-(2.94E+01) 5.06E+01-(1.64E+01) 3.72E+02-(6.23E+01) 4.93E-05-(4.25E-05) 6.82E-06-(6.54E-06) 7.00E-09(1.03E-08)
    CI+MS T1 4.72E+00-(5.49E-01) 4.45E-02-(1.96E-01) 7.15E+00-(6.92E-01) 3.15E+00-(4.27E-01) 4.72E+00-(2.90E-01) 1.20E+00-(8.22E-01) 3.97E+00-(2.64E-01) 4.40E-06-(5.17E-06) 2.72E-07-(1.33E-07) 8.55E-09(1.03E-08)
    T2 2.12E+02-(6.29E+01) 1.50E-01-(6.67E-01) 4.69E+02-(6.56E+01) 1.69E+02-(3.10E+01) 3.24E+02-(3.90E+01) 5.18E+01-(1.84E+01) 3.48E+02-(3.78E+01) 3.12E-08-(6.43E-08) 5.75E-11-(5.31E-11) 1.91E-14(5.50E-14)
    CI+LS T1 2.02E+01+(6.46E-02) 2.12E+01-(3.10E-02) 2.02E+01+(4.03E-02) 2.10E+01+(2.07E-01) 2.11E+01≈(2.43E-01) 2.02E+01+(3.51E-01) 2.12E+01-(3.79E-02) 2.12E+01-(7.51E-02) 2.12E+01-(7.26E-02) 2.11E+01(7.46E-02)
    T2 3.71E+03+(4.93E+02) 1.11E+04-(1.62E+03) 6.62E+03-(6.30E+02) 4.28E+03+(5.11E+02) 4.56E+03+(7.23E+02) 3.41E+03+(5.99E+02) 4.49E+03+(5.60E+02) 5.96E+03-(4.11E+02) 5.82E+03-(4.37E+02) 5.60E+03(4.27E+02)
    PI+HS T1 5.81E+02-(1.17E+02) 8.27E+01-(1.67E+01) 9.09E+02-(1.33E+02) 3.17E+02-(4.88E+01) 4.26E+02-(6.73E+01) 5.40E+01+(1.91E+01) 3.93E+02-(5.11E+01) 2.65E+02+(1.91E+01) 2.60E+02+(1.83E+01) 2.66E+02(2.12E+01)
    T2 8.82E+00-(2.06E+00) 2.30E-05-(2.14E-05) 2.31E+02-(5.15E+01) 1.19E+01-(2.63E+00) 1.03E+02-(2.32E+01) 1.17E-08-(3.24E-08) 5.63E+01-(1.48E+01) 1.67E-09-(2.31E-09) 4.64E-10-(2.98E-10) 1.90E-13(1.34E-13)
    PI+MS T1 3.53E+00-(5.04E-01) 1.03E-01-(4.50E-01) 7.99E+00-(8.12E-01) 2.84E+00-(4.63E-01) 5.04E+00-(3.06E-01) 1.51E-01-(4.67E-01) 3.96E+00-(4.23E-01) 6.79E-05-(5.94E-05) 2.02E-05-(1.67E-05) 1.36E-07(2.19E-07)
    T2 6.38E+02-(1.96E+02) 7.22E+01-(2.26E+01) 1.15E+05-(4.88E+04) 5.31E+02-(1.81E+02) 1.42E+04-(4.53E+03) 2.34E+02-(4.00E+02) 4.57E+03-(2.64E+03) 8.17E+01-(9.07E+00) 7.70E+01-(1.81E+01) 6.47E+01(6.47E+01)
    PI+LS T1 2.00E+01-(1.15E-01) 7.80E-01-(7.39E-01) 1.84E+01-(4.78E+00) 3.18E+00-(4.19E-01) 5.43E+00(9.38E-01) 1.73E+00-(6.82E-01) 3.94E+00-(4.98E-01) 1.51E-05-(2.78E-05) 1.32E-06-(3.66E-06) 3.82E-07(5.23E-07)
    T2 2.11E+01-(3.29E+00) 1.12E-01-(2.73E-01) 2.01E+01-(5.58E+00) 3.15E+00-(8.29E-01) 5.75E+00-(1.35E+00) 2.64E+00-(2.37E+00) 7.39E+00-(3.45E+00) 7.32E-04-(7.20E-04) 2.58E-04-(3.22E-04) 1.59E-04(1.15E-04)
    NI+HS T1 7.49E+02-(2.68E+02) 9.53E+01-(6.81E+01) 5.15E+04-(1.99E+04) 8.04E+02-(3.93E+02) 1.40E+04-(5.39E+03) 1.12E+02-(1.09E+02) 6.18E+03-(3.61E+03) 4.34E+01-(1.09E+00) 4.28E+01-(5.16E-01) 4.22E+01(8.87E-01)
    T2 2.60E+02-(4.39E+01) 3.03E+01-(1.34E+01) 5.00E+02-(6.73E+01) 2.08E+02-(6.78E+01) 3.66E+02-(3.09E+01) 5.96E+01-(1.94E+01) 3.70E+02-(6.80E+01) 5.62E-04-(1.16E-03) 1.51E-06-(4.24E-06) 5.31E-07(1.04E-06)
    NI+MS T1 4.09E-01-(6.63E-02) 2.65E-03-(4.18E-03) 1.05E+00-(1.81E-02) 3.92E-01-(6.61E-02) 9.24E-01-(5.20E-02) 6.27E-03-(9.74E-03) 8.06E-01-(8.76E-02) 3.14E-07-(1.73E-07) 3.17E-08-(5.41E-08) 5.25E-09(5.25E-09)
    T2 2.58E+01-(3.05E+00) 3.21E+00-(1.65E+00) 2.88E+00-(2.56E+00) 1.41E+01-(2.14E+00) 2.33E+01-(5.19E+00) 8.41E+00-(6.83E+00) 2.34E+01-(5.02E+00) 1.33E+00-(7.63E-01) 9.63E-01+(6.23E-01) 1.12E+00(5.42E-01)
    NI+LS T1 6.06E+02-(9.99E+01) 1.01E+02+(2.38E+01) 8.71E+02-(1.81E+02) 3.20E+02+(4.73E+01) 4.27E+02-(5.47E+01) 1.69E+02+(6.90E+01) 4.38E+02-(8.68E+01) 2.75E+02-(1.99E+01) 2.65E+02-(1.65E+01) 2.59E+02(1.96E+01)
    T2 3.62E+03-(4.60E+02) 4.07E+03-(6.99E+02) 6.65E+03-(6.71E+02) 4.14E+03-(5.23E+02) 4.31E+03-(5.90E+02) 2.89E+03-(6.43E+02) 4.52E+03-(5.55E+02) 5.98E+03-(6.43E+02) 6.07E+03-(3.37E+02) 1.99E+03(5.38E+02)
    +/≈/- 2/0/16 1/0/17 1/0/17 3/0/15 1/1/16 4/0/14 1/0/17 1/0/17 2/0/16

     | Show Table
    DownLoad: CSV
    Table 2.  Friedman test results of AMTDE-PD and nine comparison algorithms on CEC2019-SOMTP.
    Task1 Task2
    Algorithm Average rank Algorithm Average rank
    1 AMTDE-PD 2.17 AMTDE-PD 1.67
    2 MPEFMTO 3.06 MPEFMTO 3.33
    3 MTGA 3.89 MTGA 4.22
    4 MPEF 3.94 MPEF 4.22
    5 MFDE 3.94 MFDE 4.67
    6 LDA-MFEA 5.89 LDA-MFEA 5.78
    7 MFEA 7.28 MFEA 6.56
    8 EMT-EAE 7.72 EMT-EAE 8.00
    9 SBO 8.11 SBO 7.56
    10 GMFEA 9.00 GMFEA 9.00
    statistic 51.451 46.406
    P-value 0.000 0.000

     | Show Table
    DownLoad: CSV
    Table 3.  Score values of AMTDE-PD and nine comparison algorithms on CEC2019-SOMTP.
    Problem AMTDE-PD MFEA MFDE GMFEA LDA-MFEA SBO MTGA EMT-EAE MFMP MPEFMTO
    CI+HS -17.242573 4.096979 -17.101201 30.988655 0.392154 24.409735 -13.758544 22.699937 -17.242569 -17.242572
    CI+MS -18.444439 11.652982 -18.268405 35.157126 3.248721 17.936475 -10.789895 16.396293 -18.444422 -18.444438
    CI+LS 5.920443 -23.359814 31.359196 -8.055326 -2.654138 0.696335 -22.288089 2.720678 8.073585 7.587131
    PI+HS -9.000528 4.797912 -16.276617 46.481302 -5.309010 11.321329 -17.440352 3.606440 -8.975350 -9.205126
    PI+MS -12.279223 0.540779 -11.871226 48.059094 -1.920223 10.299737 -11.619883 3.340061 -12.274124 -12.274992
    PI+LS -14.112506 37.378323 -12.989220 33.089194 -6.151046 -0.104017 -8.738217 -0.148317 -14.111795 -14.112398
    NI+HS -14.054337 0.197694 -12.396925 43.392954 -2.450153 14.897147 -10.800141 9.320940 -14.053334 -14.053844
    NI+MS -18.423680 11.916704 -16.592061 29.504499 1.720064 21.714273 -12.197161 19.171619 -18.256331 -18.557927
    NI+LS -20.483169 5.271598 -14.224810 35.217437 -4.210461 1.822962 -18.571801 3.434995 5.706511 6.036738

     | Show Table
    DownLoad: CSV

    In Table 1, AMTDE-PD significantly outperforms the other nine compared algorithms in terms of average objective values on 13 out of the 18 test tasks of the CEC2019-SOMTP. According to the numbers of "+/≈/-" in Table 1, we can see that AMTDE-PD better than MFEA, MFDE, GMFEA, LDA-MFEA, SBO, MTGA, EMT-EAE, MPEF, and MPEFMTO on the CEC2019-SOMTP test suite is 16, 17, 17, 15, 16, 14, 17, 17, and 16, respectively. It can be concluded that the AMTDE-PD algorithm outperforms other algorithms on the CEC2019-SOMTP benchmark based on the accuracy of the solution and the stability of the algorithm.

    From Table 2, it can be seen that AMTDE-PD obtains the best ranks on both Task T1 and Task T2 in all test problems compared to MFEA, MFDE, GMFEA, LDA-MFEA, SBO, MTGA, EMT-EAE, MFMP, and MPEFMTO. That is to say, AMTDE-PD has the best overall performance on Task T1 and Task T2 for all the test problems. Moreover, the P-value values of AMTDE-PD on Task T1 and Task T2 are both 0, which indicates that the performance of the AMTDE-PD algorithm is significantly different from the other compared algorithms, and the AMTDE-PD algorithm outperforms the other compared algorithms in a statistically significant way.

    Table 3 shows that the AMTDE-PD achieved the best Score values on six of the nine test problems. For the complete intersection (CI) problem AMTDE-PD took the best performance scores on medium to high similarity problems and performed worse than algorithms (MFEA, GMFEA, LDA-MFEA, SBO, MTGA, EMT-EAE) using chromosome crossover for gene transfer on low similarity problems, but better than algorithms (MFDE, MFMP, MPEFMTO) using the mutation mechanism of DE for inter-task knowledge transfer. For the PI problem, AMTDE-PD obtains better Score values than the other compared algorithms on the low and medium similarity problems (PI+MS, PI+LS), and slightly worse than MTGA and MPEFMTO on the high similarity problem (PI+HS) problem. For the NI problem, AMTDE-PD obtains the best Score values on both the high similarity and the low similarity problems (NI+HS, NI+LS) achieves the best Score values and only slightly underperforms the MPEFMTO algorithm on the medium similarity problem (NI+MS). In conclusion, the proposed algorithm obtains good Score values on nine benchmark problems, and the overall performance outperforms other algorithms. This is because when the global optimum of a co-optimization task varies greatly, using either the elite solution or a randomly selected solution as the migrated knowledge between tasks may be less effective. On the contrary, AMTDE-PD uses a transfer content selection strategy based on population information, which is useful when the global optimums of the tasks differ greatly. Therefore, AMTDE-PD achieves better results on most PI and NI problems.

    To analyze the convergence behavior of AMTDE-PD and the comparison algorithms more intuitively, their convergence curves are given in Figure 4. In Figure 4, the convergence speed of AMTDE-PD is faster than that of other algorithms on most problems in CEC2019-SOMTP. However, it is slower than the algorithm for gene transfer with chromosome crossover on the CI+LS problem and slightly slower than MFDE on the NI+LS task 1 (Rastrigin function). Rastrigin function has multiple local maxima and minima, which tends to make the algorithm fall into a local optimum. MFDE generates offspring by randomly combining parents, which can obtain widely distributed solutions and escape from the local optimum. Therefore, MFDE can get better performance on the Rastrigin function.

    Figure 4.  Average convergence curves obtained by AMTDE-PD and compared algorithms On CEC2019-SOMTP.

    In summary, the AMTDE-PD algorithm has better solution accuracy and faster convergence speed than other algorithms, can solve the single-objective MFO problem effectively and demonstrates a more competitive and comprehensive performance than other EMTO algorithms.

    In order to evaluate the convergence of the algorithms more fairly, the maximum number of function evaluations of all the algorithms was increased to 200,000 times for the experiments, and the convergence curves of AMTDE-PD and its comparison algorithms are shown in Figure 5. From Figure 5, we can see that with sufficient function evaluation times, AMDE has competitiveness in terms of convergence speed.

    Figure 5.  Average convergence curves obtained by AMTDE-PD and compared algorithms On CEC2019-SOMTP for maximum function evaluation as 200,000.

    Compare AMTDE-PD with two classical evolutionary multitasking optimization algorithms, MFEA and MFDE, and two recent high-performance evolutionary multitasking optimization algorithms, MPEF, and MPEFMTO, on WCCI2020-SOCTP. The reason for choosing these comparison algorithms is that these algorithms use different transfer content selection strategies. Among them, MFEA and MFDE use the randomized solutions in the task as the transfer content. MPEF uses the difference vector of the randomized solutions in the task and the elite solutions as the transfer content. The transfer content in MPEFMTO is the dominant solution, the inferior solution and the difference vector of the randomized solutions in the task. The average objective function values obtained from 20 independent runs of each algorithm on each task and the standard deviation were recorded as results. The results are shown in Table 4, and the results of the Wilcoxon rank sum test with a significance level α = 0.05 are given in the last row. The synergy performance metric Score is also utilized to verify the comprehensive performance of each algorithm on each problem containing two complex tasks. The values of the performance metric Score for the AMTDE-PD, MFEA, MPDE, MFMP, and MPEFMTO are given in Table 5.

    Table 4.  Experimental results of AMTDE-PD algorithm and other nine comparison algorithms on WCCI2020-SOCTP test suite.
    Problem Task MFEA MFDE MPEF MPEFMTO AMTDE-PD
    P1 Task1 6.4841E+02(4.8164e+00)- 6.0408E+02(2.5657e+00)- 6.0084E+02(5.6434E-01)- 6.0076E+02(4.4438E-01)- 6.0059E+02(3.8116e-01)
    Task2 6.4832E+02(4.7256e+00)- 6.0442E+02(2.5177e+00)- 6.0101E+02(5.6434E-01)- 6.0094E+02(5.2927E-01)- 6.0067E+02(3.1623e-01)
    P2 Task1 7.0107E+02(2.5362e-02)- 7.0000E+02(7.0583e-05)≈ 7.0000E+02(7.0972E-06)≈ 7.0000E+02(1.6550E-03)≈ 7.0000E+02(1.6538e-03)
    Task2 7.0107E+02(1.4481e-02)- 7.0000E+02(3.8944e-03)≈ 7.0000E+02(1.7538E-06)≈ 7.0000E+02(2.3784E-03)≈ 7.0000E+02(2.2084e-03)
    P3 Task1 3.0247E+06(1.6196e+06)- 6.3359E+06(2.0626e+06)- 8.9800E+03(4.3220E+03)+ 1.3640E+04(1.6415E+04)+ 2.1596E+04(3.9472e+04)
    Task2 2.9191E+06(1.2221e+06)- 6.4765E+06(2.9946e+06)- 9.4501E+03(3.5152E+03)- 1.0963E+04(6.9444E+03)- 9.1345E+03(2.8387e+03)
    P4 Task1 1.3006E+03(8.7430e-02)- 1.3006E+03(8.0022e-02)- 1.3004E+03(4.4541E-02)≈ 1.3004E+03(3.6728E-02)≈ 1.3004E+03(4.8408e-02)
    Task2 1.3005E+03(6.5949e-02)- 1.3005E+03(7.0733e-02)- 1.3003E+03(5.1003E-02)≈ 1.3003E+03(4.2129E-02)≈ 1.3003E+03(4.5993e-02)
    P5 Task1 1.5611E+03(1.3037e+01)- 1.5330E+03(2.0512e+00)- 1.5222E+03(1.8118E+00)+ 1.5225E+03(1.8091E+00)+ 1.5226E+03(1.8020E+00)
    Task2 1.5523E+03(1.2078e+01)- 1.5339E+03(1.7193e+00)- 1.5237E+03(1.8002E+00)≈ 1.5235E+03(1.4926E+00)+ 1.5237E+03(1.6514E+00)
    P6 Task1 1.8249E+06(7.9999e+05)- 2.5073E+06(1.1612e+06)- 1.6597E+04(2.7983E+04)+ 1.2602E+04(7.2662E+03)+ 3.1416E+04(8.6112E+04)
    Task2 1.8805E+06(8.9531e+05)- 1.9593E+06(7.6951e+05)- 1.0719E+04(8.4671E+03)- 9.6431E+03(4.8180E+03)- 8.6457E+03(1.5279E+03)
    P7 Task1 3.3196E+03(3.5229e+02)- 3.8522E+03(1.7922e+02)- 2.5376E+03(1.1344E+02)- 2.5312E+03(1.0284E+02)- 2.5238E+03(1.1907E+02)
    Task2 3.3172E+03(3.1922e+02)- 3.9439E+03(1.9536e+02)- 2.6351E+03(1.1284E+02)- 2.6156E+03(1.5219E+02)+ 2.6235e+03(1.1527e+02)
    P8 Task1 5.2024E+02(7.6601e-02)+ 5.2120E+02(3.9190e-02)- 5.2119E+02(5.1409E-02)- 5.2118E+02(4.4255E-02)≈ 5.2118E+02(6.1730E-02)
    Task2 5.2029E+02(7.9562e-02)+ 5.2121E+02(3.5954e-02)- 5.2121E+02(3.0681E-02)- 5.2119E+02(4.1841E-02)- 5.2118E+02(4.0389E-02)
    P9 Task1 8.2322E+03(1.2798e+03)+ 1.4742E+04(4.0996e+02)- 1.1287E+04(4.1119E+02)- 1.1396E+04(5.0014E+02)- 1.1261E+04(4.6353e+02)
    Task2 1.6217E+03(5.4480e-01)- 1.6227E+03(1.8758e-01)- 1.6215E+03(3.5049E-01)- 1.6214E+03(3.8833E-01)≈ 1.6214e+03(4.2650e-01)
    P10 Task1 3.2613E+04(1.5889e+04)- 2.7934E+04(1.0828e+04)- 2.4966E+03(1.2714E+03)- 2.2924E+03(2.5205E+02)- 2.2409E+03(4.0205E+01)
    Task2 3.0951E+06(2.7513e+06)- 2.3497E+06(8.6756e+05)- 5.0067E+04(1.2795E+05)- 1.7257E+04(1.4961E+04)+ 1.7769E+04(407316E+04)
    +/≈/- 3/0/17 0/2/18 3/5/12 6/6/8

     | Show Table
    DownLoad: CSV
    Table 5.  The Score values of performance metrics for AMTDE-PD, MFEA, MPDE, MFMP and MPEFMTO.
    Problem AMTDE-PD MFEA MFDE MPEF MPEFMTO
    P1 -9.91003 35.63654 -6.45095 -9.61851 -9.65704
    P2 -8.94667 35.77683 -8.92343 -8.9653 -8.94144
    P3 -12.9939 9.08291 29.5785 -12.95587 -12.93474
    P4 -11.70563 16.20448 22.44774 -13.16699 -13.7796
    P5 -11.99945 32.42007 4.20295 -12.21723 -12.40634
    P6 -13.7214 17.81398 23.55646 -13.7802 -13.86884
    P7 -13.33472 10.20006 29.38105 -13.03279 -13.2136
    P8 8.31523 -35.60554 9.60732 9.2293 8.4537
    P9 -4.06042 -13.22761 28.56233 -5.82066 -5.45363
    P10 -13.6617 21.68881 19.39499193 -13.17343 -13.65862

     | Show Table
    DownLoad: CSV

    As can be seen in Table 4, AMTDE-PD outperforms MFEA, MFDE, MPEF, and MPEFMTO in the CCI2020-SOCTP test suite on 17, 18, and 12 tasks, respectively. However, MFEA, MFDE, MPEF, and MPEFMTO approximate AMTDE-PD on 0, 2, 5, and 6 tasks, respectively. In addition, the synergy performance metric scores of the algorithms in Table 5 show that the overall performance of AMTDE-PD is superior to that of MFEA, MFDE, MPEF and MPEFMTO on 8, 10, 6, and 6 problems, respectively. The above statistical results show that AMTDE-PD can effectively deal with complex multitasking optimization problems.

    To analyze the effectiveness of the inter-task transfer strategy used in AMTDE-PD, we define three variants of AMTDE-PD. One is No-AMTDE-PD with no inter-task knowledge transfer, another is Rand-AMTDE-PD with randomized solutions transfer between tasks, and another is Elist-AMTDE-PD with elite solutions transfer between tasks. AMTDE-PD is compared with these three variants on CEC2019-SOMTP, with each variant running 20 independent runs. The comparison results are shown in Table 6. In Table 6, it is clear that Rand-AMTDE-PD, Elist-AMTDE-PD, and AMTDE-PD outperform No-AMTDE-PD on most of the tasks, which indicates that inter-task knowledge transfer can facilitate co-optimization of tasks. Among them, Elist-AMTDE-PD performs better than No-AMTDE-PD on high and medium similarity problems and worse than No-AMTDE-PD on some low similarity problems. This is because if the tasks have high similarity, transferring the elite solutions between the tasks can effectively improve the co-optimization of the tasks. On the contrary, if the tasks are not similar or have low similarity, negative transfer will occur, thus affecting the performance of the algorithm. Rand-AMTDE-PD outperforms No-AMTDE-PD on 13 tasks and underperforms No-AMTDE-PD on 5. The above phenomenon is attributed to the random selection of what to transfer between tasks, which maintains population diversity to some extent due to the random nature of the transfer, but is not conducive to the full utilization of useful information between tasks. AMTDE-PD outperforms No-AMTDE-PD, Elist-AMTDE-PD, and Rand-AMTDE-PD on 15, 16, and 14 tasks out of the 18 tasks, respectively. The results show that the transfer strategy based on population distribution information proposed by us can effectively promote collaborative optimization but cannot completely avoid negative transfer between tasks, especially for problems with no intersection and low similarity.

    Table 6.  Comparison of solution accuracy of AMTDE-PD with Rand-AMTDE-PD, Elist-AMTDE-PD and No-AMTDE-PD.
    Problem Task No-AMTDE-PD Rand-AMTDE-PD Elist-AMTDE-PD AMTDE-PD
    CI+HS T1 1.43E-09 6.55E-09 3.34E-10 4.80E-12
    T2 2.50E+02 1.99E-05 5.73E-06 7.00E-09
    CI+MS T1 3.27E-07 1.57E-08 5.08E-07 8.55E-09
    T2 2.50E+02 1.41E-13 2.90E-10 1.91E-14
    CI+LS T1 2.11E+01 2.12E+01 2.12E+01 2.11E+01
    T2 4.94E+03 5.61E+03 6.31E+03 5.60E+03
    PI+HS T1 2.68E+02 2.67E+02 2.48E+02 2.66E+02
    T2 2.04E-09 1.92E-13 1.21E-13 1.90E-13
    PI+MS T1 2.90E-07 8.17E-08 2.09E-01 1.36E-07
    T2 4.87E+01 4.90E+01 4.86E+01 6.47E+01
    PI+LS T1 5.36E-07 2.85E-07 1.30E-06 3.82E-07
    T2 1.43E-03 2.42E-04 4.49E-04 1.59E-04
    NI+HS T1 4.79E+01 4.23E+01 4.30E+01 4.22E+01
    T2 2.53E+02 1.12E-03 5.85E-01 5.31E-07
    NI+MS T1 3.09E-05 6.41E-07 6.18E-09 5.25E-09
    T2 1.15E+00 7.44E-01 5.56E+00 1.12E+00
    NI+LS T1 2.39E+02 2.61E+02 2.67E+02 2.59E+02
    T2 5.23E+03 1.77E+03 8.29E+03 1.99E+03

     | Show Table
    DownLoad: CSV

    In AMTDE-PD, there are two main control parameters δ and q, which δ controls the update condition of the RMP and q are used to adjust the value of the RMP. To analyze the effect of parameter q on the AMTDE-PD algorithm, we compare the comprehensive performance of AMTDE-PD with different values of parameter q on all problems. In this experiment, set q=0.4,q=0.5,q=0.6,q=0.7,q=0.8,q=0.9. The results of Friedman's test with different parameter settings are shown in Table 7. It can be seen from Table 7 that the parameter q has no significant effect on the performance of AMTDE-PD when all tasks are considered. However, from the average rank, AMTDE-PD has the best rank on T2 and has the second rank on T1 for the case of q = 0.9. So, in this paper, we recommend q = 0.9.

    Table 7.  Friedman test results of AMTDE-PD with different parameter settings.
    Task1 Task2
    Parameter Rank Parameter Rank
    q = 0.9 3.11 q = 0.9 2.67
    q = 0.8 2.67 q = 0.8 3.67
    q = 0.7 3.78 q = 0.7 3.00
    q = 0.6 4.33 q = 0.6 2.78
    q = 0.5 3.89 q = 0.5 4.76
    q = 0.4 3.22 q = 0.4 4.22
    statistic 5 statistic 5
    P value 0.448 P value 0.122

     | Show Table
    DownLoad: CSV

    To analyze the influence of parameter δ on the performance of AMTDE-PD, we compared the performance of AMTDE-PD with different values of δ on all the problems in the single-target classic test set. The Friedman test for the comparison results is shown in Table 8. As shown in Table 8, the P-values for tasks T1 and T2 on all test problems are 0.216 and 0.463, respectively, both of which are greater than the significance level of 0.05, indicating that the value of δ has no significant impact on the performance of the algorithm AMTDE-PD. However, from the average ranking, δ = 0.6 achieved the best ranking on task T1, followed by δ = 0.5. The best ranking was obtained with δ = 0.5 on task T2, followed by δ = 0.3, δ = 0.6, δ = 0.4, and δ = 0.2. Through the above analysis, δ = 0.5 is recommended.

    Table 8.  Friedman test results of AMTDE-PD with different δ values.
    T1 T2
    Parameter Rank Parameter Rank
    δ=0.2 3.78 δ=0.2 3.44
    δ=0.3 3.28 δ=0.3 2.83
    δ=0.4 2.94 δ=0.4 3.39
    δ=0.5 2.83 δ=0.5 2.22
    δ=0.6 2.17 δ=0.6 3.11
    statistic 5.272 statistic 3.598
    P value 0.216 P value 0.463

     | Show Table
    DownLoad: CSV

    To validate the effectiveness of the population similarity-based RMP in AMTDE-PD, we compare the proposed similarity-based RMP (SIM-RMP) with the pre-improvement RMP on CI+HS, PI+MS, NI+LS, CI+LS, PI+LS, and NI+ML problems are compared. The comparison results are shown in Figure 6, from which we can see that SIM-RMP has larger values for the CI+HS problem and smaller values for the PIMS and NI+LS problems.

    Figure 6.  Comparison of random interaction probability before and after improvement.

    That is to say, the values of the proposed SIM-RMP on different problems change significantly as the evolutionary algebra advances, while the RMP values before the improvement change insignificantly but oscillate significantly. This is because the improvement rate of the solution is constantly changing during the evolutionary process, especially for problems with many local optimal solutions, the improvement rate of the solution changes more frequently during the evolutionary process. Thus, adjusting the RMP based only on the improvement rate of the solution during the evolutionary process makes it take values that oscillate sharply. In addition, it can also be seen from Figure 6 that the SIM-RMP proposed in this paper has a larger value in the CI+LS problem, while the RMP before improvement has a smaller value. Therefore, the proposed SIM-RMP can effectively capture the correlation between two tasks, which in turn can effectively control the intensity of inter-task interactions, increase the efficiency of inter-task knowledge transfer, and enhance the performance of the algorithm.

    In order to make full use of the valuable information between tasks and reduce the negative transfer, we propose an adaptive multitasking differential optimization algorithm based on population distribution information (AMTDE-PD). Specifically, first, the differences in task population distribution are utilized to determine the valuable information transferred between tasks. Next, using population distribution information to evaluate the evolutionary trend between tasks and adjusting the interaction probability between tasks based on the evolutionary trend, the adaptive adjustment of interaction intensity is achieved to reduce negative transfer between tasks. Then, DE algorithm is used as the task solver to solve the related tasks. The AMTDE-PD algorithm proposed in this paper and other mainstream evolutionary multitasking optimization algorithms are experimented on CEC2019-SOMTP and WCCI2020-SOCTP, respectively, and the experimental results show that the proposed AMTDE-PD algorithm is able to solve different types of multitasking optimization problems effectively and has high robustness.

    We declare that we have not used Artificial Intelligence (AI) tools in the creation of this article.

    This work was supported by National Natural Science Foundation of China under Grant (No. 62176146, No. 62272384), National Social Science Foundation of China under Grant (No. 21XTY012), National Education Science Foundation of China under Grant (No. BCA200083), Key Project of Shaanxi Provincial Natural Science Basic Research Program under Grant (No. 2023-JC-ZD-34), and Natural Science Basic Research Program of Shaanxi under Grant (No. 2022JM-050).

    The authors declare that they have no conflicts of interest.



    [1] Aikman D, Galesic M, Gigerenzer G, et al. (2014) Taking uncertainty seriously: simplicity versus complexity in financial regulation. Ind Corp Change 30: 317–345. https://doi.org/10.1093/icc/dtaa024 doi: 10.1093/icc/dtaa024
    [2] Allan N, Cantle N, Godfrey P (2013) A review of the use of complex systems applied to risk appetite and emerging risks in ERM practice: Recommendations for practical tools to help risk professionals tackle the problem of risk appetite and emerging risks. Brit Actuar J 18: 163–234. https://doi.org/10.1017/S135732171200030X doi: 10.1017/S135732171200030X
    [3] Allen F, Gale D (2000) Financial Contagion. J Polit Econ 108: 1–33. https://doi.org/10.1086/262109 doi: 10.1086/262109
    [4] Althaus CE (2005) A disciplinary perspective on the epistemological status of risk. Risk Anal 25: 567–588. https://doi.org/10.1111/j.1539-6924.2005.00625.x doi: 10.1111/j.1539-6924.2005.00625.x
    [5] Anagnostou I, Sourabh S, Kandhai D (2018) Incorporating Contagion in Portfolio Credit Risk Models using Network Theory. Complexity 2018. https://doi.org/10.1155/2018/6076173 doi: 10.1155/2018/6076173
    [6] Anderson N, Noss J (2013) The Fractal Market Hypothesis and its implications for the stability of financial markets. Available from: https://www.bankofengland.co.uk/-/media/boe/files/financial-stability-paper/2013/the-fractal-market-hypothesis-and-its-implications-for-the-stability-of-financial-markets.pdf?la=en&hash=D096065EA97BE61902BDE5CC022D1E6EA38AAC49.
    [7] Anderson PW (1972) More is different. Science 177: 393–396. https://doi.org/10.1126/science.177.4047.393 doi: 10.1126/science.177.4047.393
    [8] Arthur WB (2021) Foundations of Complexity. Nat Rev Phys 3: 136–145. https://doi.org/10.1038/s42254-020-00273-3 doi: 10.1038/s42254-020-00273-3
    [9] Arthur WB (2013) Complexity Economics: A Different Framework for Economic Thought. Complexity Economics, Oxford University Press. https://doi.org/10.2469/dig.v43.n4.70
    [10] Arthur WB (1999) Complexity and the economy. Science 284: 107–09. https://doi.org/10.1126/science.284.5411.107 doi: 10.1126/science.284.5411.107
    [11] Aven T (2017) A Conceptual Foundation for Assessing and Managing Risk, Surprises and Black Swans. In: Motet, G., Bieder, C., The Illusion of Risk Control, Springer Briefs in Applied Sciences and Technology. Springer, Cham. https://doi.org/10.1007/978-3-319-32939-0_3
    [12] Aven T (2013) On how to deal with deep uncertainties in risk assessment and management context. Risk Anal 33: 2082–2091. https://doi.org/10.1111/risa.12067 doi: 10.1111/risa.12067
    [13] Aven T, Flage R (2015) Emerging risk – Conceptual definition and a relation to black swan type of events. Reliab Eng Syst Safe 144: 61–67. https://doi.org/10.1016/j.ress.2015.07.008 doi: 10.1016/j.ress.2015.07.008
    [14] Aven T, Krohn BS (2014) A new perspective on how to understand, assess and manage risk and the unforeseen. Reliab Eng Syst Safe 121: 1–10. https://doi.org/10.1016/j.ress.2013.07.005 doi: 10.1016/j.ress.2013.07.005
    [15] Babbie ER (2011) The Practice of Social Science Research. South African Ed. Cape Town. Oxford University Press.
    [16] Baranger M (2010) Chaos, Complexity, and Entropy. A Physics Talk for Non-Physicists. Report in New England Complex Systems Institute, Cambridge.
    [17] Barberis N (2017) Behavioural Finance: Asset Prices and Investor Behaviour. Available from: https://www.aeaweb.org/content/file?id=2978.
    [18] Basel Committee on Banking Supervision (2017) Basel Ⅲ: Finalising Post-Crisis Reforms. Available from: https://www.bis.org/bcbs/publ/d424.htm.
    [19] Basel Committee on Banking Supervision (2013) Fundamental Review of the Trading Book: A Revised Market Risk Framework. Available from: https://www.bis.org/publ/bcbs265.pdf.
    [20] Basel Committee on Banking Supervision (2006) Basel Ⅱ International Convergence of Capital Measurement and Capital Standards: A Revised Framework-Comprehensive Version. Available from: https://www.bis.org/publ/bcbs128.pdf.
    [21] Basel Committee on Banking Supervision (1996) Amendment to the Capital Accord to Incorporate Market Risks. Available from: https://www.bis.org/publ/bcbs24.pdf.
    [22] Basel Committee on Banking Supervision (BCBS) (1988) International Convergence of Capital Measurement and Capital Standards. Available from: https://www.bis.org/publ/bcbs107.pdf.
    [23] Battiston S, Caldarelli G, May R, et al. (2016a) The Price of Complexity in Financial Networks. P Natl A Sci 113: 10031–10036. https://doi.org/10.1073/pnas.1521573113. doi: 10.1073/pnas.1521573113
    [24] Battiston S, Farmer JD, Flache A, et al. (2016b) Complexity theory and financial regulation: Economic policy needs interdisciplinary network analysis and behavioural modelling. Science 351: 818–819. https://doi.org/10.1126/science.aad0299 doi: 10.1126/science.aad0299
    [25] Bayes T, Price R (1763) An Essay towards Solving a Problem in the Doctrine of Chances. Philos T R Soc London 53: 370–418. https://doi.org/10.1098/rstl.1763.0053 doi: 10.1098/rstl.1763.0053
    [26] Beck U (1992) Risk society: towards a new modernity. Sage London.
    [27] Beissner P, Riedel F (2016) Knight-Walras Equilibria. Centre for Mathematical Economics Working Paper 558.
    [28] Bharathy GK, McShane MK (2014) Applying a Systems Model to Enterprise Risk Management. Eng Manag J 26: 38–46. https://doi.org/10.1080/10429247.2014.11432027 doi: 10.1080/10429247.2014.11432027
    [29] Black F, Scholes M (1973) The Pricing of Options and Corporate Liabilities. J Polit Econ 29: 449–470.
    [30] Blume L, Durlauf (2006) The Economy as an Evolving Complex System Ⅲ: current perspectives and future directios. https://doi.org/10.1093/acprof:oso/9780195162592.001.0001
    [31] Bolton P, Despres M, Pereira da Silva L, et al. (2020) The green swan: central banking and financial stability in the age of climate change. Available from: https://www.bis.org/publ/othp31.pdf.
    [32] Bonabeau E (2002) Agent-Based Modeling: Methods and Techniques for Simulating Human Systems. Natl Acad Sci 99: 7280–7287. https://doi.org/10.1073/pnas.082080899 doi: 10.1073/pnas.082080899
    [33] Brace I (2008) Questionnaire Design: How to Plan, Structure, and Write Survey Material for Effective Market Research, 2 Eds., London: Kogan Page. https://doi.org/10.5860/choice.42-3520
    [34] Bronk R (2011) Uncertainty, modeling monocultures and the financial crisis. Bus Econ 42: 5–18.
    [35] Bruno B, Faggini M, Parziale A (2016) Complexity Modeling in Economics: the state of the Art. Econ Thought 5: 29–43.
    [36] Burzoni M, Riedel F, Soner MH (2021) Viability and arbitrage under Knightian uncertainty. Econometrica 89: 1207–1234. https://doi.org/10.3982/ECTA16535 doi: 10.3982/ECTA16535
    [37] Chan-Lau JA (2017) An Agent Based Model of the Banking System. IMF Working Paper. https://doi.org/10.5089/9781484300688.001
    [38] Chaudhuri A, Ghosh SK (2016) Quantitative Modeling of Operational Risk in Finance and Banking Using Possibility Theory, Springer International Publishing Switzerland. https://doi.org/10.1007/978-3-319-26039-6
    [39] Collis J, Hussey R (2009) Business Research: a practical guide for undergraduate and postgraduate students, 3rd ed., New York: Palgrave Macmillan.
    [40] Corrigan J, Luraschi P, Cantle N (2013) Operational Risk Modeling Framework. Milliman Research Report. Available from: https://web.actuaries.ie/sites/default/files/erm-resources/operational_risk_modelling_framework.pdf.
    [41] De Bondt W, Thaler R (1985) Does the stock market overact? J Financ 40: 793–808. https://doi.org/10.2307/2327804 doi: 10.2307/2327804
    [42] Diebold FX, Doherty NA, Herring RJ (2010) The Known, Unknown, and the Unknowable. Princeton, NJ: Princeton University Press.
    [43] Dorigo M (2007) Editorial. Swarm Intell 1: 1–2. https://doi.org/10.1007/s11721-007-0003-z doi: 10.1007/s11721-007-0003-z
    [44] Dorigo M, Maniezzo V, Colorni A (1996) Ant system: optimisation by a colony cooperating agents. IEEE T Syst Man Cy B 26: 29–41. https://doi.org/10.1109/3477.484436 doi: 10.1109/3477.484436
    [45] Dowd K (1996) The case for financial Laissez-faire. Econ J 106: 679–687. https://doi.org/10.2307/2235576 doi: 10.2307/2235576
    [46] Dowd K, Hutchinson M (2014) How should financial markets be regulated? Cato J 34: 353–388.
    [47] Dowd K, Hutchinson M, Ashby S, et al. (2011) Capital Inadequacies the Dismal Failure of the Basel Regime of Bank Capital Regulation. Banking & Insurance Journal.
    [48] Easterby-Smith M, Thorpe R, Jackson PR (2015) Management and Business Research, 5 Eds., London: Sage.
    [49] Ellinas C, Allan N, Combe C (2018) Evaluating the role of risk networks on risk identification, classification, and emergence. J Network Theory Financ 3: 1–24. https://doi.org/10.21314/JNTF.2017.032 doi: 10.21314/JNTF.2017.032
    [50] Ellsberg D (1961) Risk, ambiguity, and the savage axioms. Q J Econ 75: 643–669. https://doi.org/10.2307/1884324 doi: 10.2307/1884324
    [51] Evans JR, Allan N, Cantle N (2017) A New Insight into the World Economic Forum Global Risks. Econ Pap 36. https://doi.org/10.1111/1759-3441.12172 doi: 10.1111/1759-3441.12172
    [52] Fadina T, Schmidt T (2019) Default Ambiguity. Risks 7: 64. https://doi.org/10.3390/risks7020064 doi: 10.3390/risks7020064
    [53] Fama EF (1970) Efficient Capital Markets: A review of Theory and Empirical Work. J Financ 25: 383–417. https://doi.org/10.2307/2325486 doi: 10.2307/2325486
    [54] Farmer JD (2012) Economics needs to treat the economy as a complex system. Institute for New Economic Thinking at Oxford Martin School. Working Paper.
    [55] Farmer D (2002) Market force, ecology, and evolution. Ind Corp Change 11: 895–953. https://doi.org/10.1093/icc/11.5.895 doi: 10.1093/icc/11.5.895
    [56] Farmer D, Lo A (1999) Fronties of finance: Evolution and efficient markets. Pro Nat Acad Sci 96: 9991–9992. https://doi.org/10.1073/pnas.96.18.99 doi: 10.1073/pnas.96.18.99
    [57] Farmer JD, Gallegati C, Hommes C, et al. (2012) A complex systems approach to constructing better models for financial markets and the economy. Eur Phys J-Spec Top 214: 295–324. https://link.springer.com/article/10.1140/epjst/e2012-01696-9
    [58] Farmer JD, Foley D (2009) The economy needs agent-based modeling. Nature 460: 685–686. https://www.nature.com/articles/460685a
    [59] Fellegi IP (2010) Survey Methods and Practices, Available from: https://unstats.un.org/wiki/download/attachments/101354131/12-587-x2003001-eng.pdf?api=v2.
    [60] Field A (2009) Discovering Statistics Using SPSS, 3 eds., London: Sage.
    [61] Fink A (2009) How to Conduct Surveys, 4 eds., Thousand Oaks, CA: Sage.
    [62] Fink A (1995) How to Measure Survey Reliability and Validity, Thousand Oaks, CA: Sage.
    [63] Forrester JW (1969) Urban Dynamics. Cambridge Mass M.I.T Press.
    [64] Ganegoda A, Evans J (2014) A framework to manage the measurable, immeasurable and the unidentifiable financial risk. Aust J Manage 39: 5–34. https://doi.org/10.1177/0312896212461033 doi: 10.1177/0312896212461033
    [65] Gao Q, Fan H, Shen J (2018) The Stability of Banking System Based on Network Structure: An Overview. J Math Financ 8: 517–526. 10.4236/jmf.2018.83032 doi: 10.4236/jmf.2018.83032
    [66] Gebizlioglu OL, Dhaene J (2009) Risk Measures and Solvency-Special Issue. J Comput Appl Math 233: 1–2. https://doi.org/10.1016/j.cam.2009.07.030 doi: 10.1016/j.cam.2009.07.030
    [67] Gigerenzer G (2007) Gut feelings: The Intelligence of the unconscious. New York: Viking Press.
    [68] Gigerenzer G, Hertwig R, Pachur T (Eds) (2011) Heuristics: The foundations of adaptive behaviour, New York: Oxford University Press.
    [69] Glasserman P, Young HP (2015) How likely is contagion in financial networks? J Bank Financ 50: 383–399. https://doi.org/10.1016/j.jbankfin.2014.02.006 doi: 10.1016/j.jbankfin.2014.02.006
    [70] Gros S (2011) Complex systems and risk management. IEEE Eng Manag Rev 39: 61–72.
    [71] Hair J, Black W, Babin B, et al (2014) Multivariate Data Analysis, 7 eds., Pearson Prentice Hall, Auflage.
    [72] Haldane AG (2017) Rethinking Financial Stability. Available from: https://www.bis.org/review/r171013f.pdf.
    [73] Haldane AG (2016) The dappled world. Available from: https://www.bankofengland.co.uk/-/media/boe/files/speech/2016/the-dappled-world.pdf.
    [74] Haldane AG (2012) The Dog and the Frisbee. Available from: https://www.bis.org/review/r120905a.pdf.
    [75] Haldane AG (2009) Rethinking the financial network. Available from: https://www.bis.org/review/r090505e.pdf.
    [76] Haldane AG, May RM (2011) Systemic risk in banking ecosystems. Available from: https://www.nature.com/articles/nature09659.
    [77] Hayek FA (1964) The theory of complex phenomena. The critical approach to science and philosophy, 332–349. https://doi.org/10.4324/9781351313087-22
    [78] Helbing D (2013) Globally network risks and how to respond. Nature 497: 51–59. https://doi.org/10.1038/nature12047 doi: 10.1038/nature12047
    [79] Holland JH (2002) Complex adaptive systems and spontaneous emergence. In Curzio, A.Q., Fortis, M., Complexity and Industrial Clusters: Dynamics and Models in Theory and Practice, New York: Physica-Verlag, 25–34. https://doi.org/10.1007/978-3-642-50007-7_3
    [80] Hommes C (2011) The heterogeneous expectations hypothesis: Some evidence from the lab. J Econ Dyn Control 35: 1–24. https://doi.org/10.1016/j.jedc.2010.10.003 doi: 10.1016/j.jedc.2010.10.003
    [81] Hommes C (2001) Financial markets as nonlinear adaptive evolutionary systems. Quant Financ 1: 149–167. https://doi.org/10.1080/713665542 doi: 10.1080/713665542
    [82] Hull J (2018) Risk Management and Financial Institutions, 5 eds., John Wiley.
    [83] Ingram D, Thompson M (2011) Changing seasons of risk attitudes. Actuary 8: 20–24.
    [84] Joshi D (2014) Fractals, liquidity, and a trading model. European Investment Strategy, BCA Research.
    [85] Karp A, van Vuuren G (2019) Investment's implications of the Fractal Market Hypothesis. Ann Financ Econ 14: 1–27. https://doi.org/10.1142/S2010495219500015 doi: 10.1142/S2010495219500015
    [86] Katsikopoulos KV (2011) Psychological Heuristics for Making Inferences: Definition, Performance, and the Emerging Theory and Practice. Decis Anal 8: 10–29. https://doi.org/10.1287/deca.1100.0191 doi: 10.1287/deca.1100.0191
    [87] Keynes JM (1921) A Treatise on Probability, London: Macmillan.
    [88] Kimberlin CL, Winterstein AG (2008) Validity and Reliability of Measurement Instruments in Research. Am J Health Syst Pharma 65: 2276–2284. https://doi.org/10.2146/ajhp070364. doi: 10.2146/ajhp070364
    [89] Kirman A (2006) Heterogeneity in economics. J Econ Interact Co-ordination 1: 89–117. https://doi.org/10.1007/s11403-006-0005-8 doi: 10.1007/s11403-006-0005-8
    [90] Kirman A (2017) The Economy as a Complex System. In: Aruka, Y., Kirman, A., Economic Foundations for Social Complexity Science. Evolutionary Economics and Social Complexity Science, Singapore. https://doi.org/10.1007/978-981-10-5705-2_1
    [91] Klioutchnikova I, Sigovaa M, Beizerova N (2017) Chaos Theory in Finance. Procedia Comput Sci 119: 368–375. https://doi.org/10.1016/j.procs.2017.11.196 doi: 10.1016/j.procs.2017.11.196
    [92] Knight FH (1921) Risk, Uncertainty and Profit, Boston and New York: Houghton Mifflin Company.
    [93] Krejcie RV, Morgan DW (1970) Determining the sample size for research activities. Educ Psychol Meas 30: 607–610.
    [94] Kremer E (2012) Modeling and Simulation of Electrical Energy Systems though a Complex Systems Approach using Agent Based Models, Kit Scientific Publishing.
    [95] Kurtz CF, Snowden DJ (2003) The new dynamics of strategy: Sense-making in a complex and complicated world. IBM Syst J 42: 462–483. https://doi.org/10.1147/sj.423.0462 doi: 10.1147/sj.423.0462
    [96] Li Y (2017) A non-linear analysis of operational risk and operational risk management in banking industry, PhD Thesis, School of Accounting, Economics and Finance, University of Wollongong. Australia.
    [97] Li Y, Allan N, Evans J (2018) An analysis of the feasibility of an extreme operational risk pool for banks. Ann Actuar Sci 13: 295–307. https://doi.org/10.1017/S1748499518000222 doi: 10.1017/S1748499518000222
    [98] Li Y, Allan N, Evans J (2017) An analysis of operational risk events in US and European banks 2008–2014. Ann Actuar Sci 11: 315–342. https://doi.org/10.1017/S1748499517000021 doi: 10.1017/S1748499517000021
    [99] Li Y, Allan N, Evans J (2017) A nonlinear analysis of operational risk events in Australian banks. J Oper Risk 12: 1–22. https://doi.org/10.21314/JOP.2017.185 doi: 10.21314/JOP.2017.185
    [100] Li Y, Evans J (2019) Analysis of financial events under an assumption of complexity. Ann Actuar Sci 13: 360–377. https://doi.org/10.1017/S1748499518000337 doi: 10.1017/S1748499518000337
    [101] Li Y, Shi L, Allan N, et al. (2019) An Analysis of power law distributions and tipping points during the global financial crisis. Ann Actuar Sci 13: 80–91. https://doi.org/10.1017/S1748499518000088 doi: 10.1017/S1748499518000088
    [102] Liu G, Song G (2012) EMH and FMH: Origin, evolution, and tendency. 2012 Fifth International Workshop on Chaos-fractals Theories and Applications. https://doi.org/10.1109/IWCFTA.2012.71
    [103] Lo A (2004) The adaptive markets hypothesis: Market efficiency from an evolutionary perspective. J Portfolio Manage 30: 15–29. https://doi.org/10.3905/jpm.2004.442611 doi: 10.3905/jpm.2004.442611
    [104] Lo A (2012) Adaptive markets and the new world order. Financ Anal J 68: 18–29. https://doi.org/10.2469/faj.v68.n2.6 doi: 10.2469/faj.v68.n2.6
    [105] Macal CM, North MJ, Samuelson MJ (2013) Agent-Based Simulation. In: Gass, S.I., Fu M.C., Encyclopaedia of Operations Research and Management Science, Springer, Boston MA.
    [106] Mandelbrot B (1963) The Variation of Certain Speculative Prices. J Bus 36: 394–419. http://www.jstor.org/stable/2350970
    [107] Mandelbrot B, Hudson RL (2004) The (Mis) behaviour of Markets: A Fractal View of Risk, Ruin, and Reward, Basic Books: Cambridge.
    [108] Mandelbrot B, Taleb N (2007) Mild vs. wild randomness: Focusing on risks that matter. In: Diebold, F.X., Doherty, N.A., Herring, R.J., The Known, the Unknown, and the Unknowable in Financial Risk Management: Measurement and Theory Advancing Practice, Princeton, NJ: Princeton University Press, 47–58.
    [109] Markowitz HM (1952) Portfolio Selection. J Financ 7: 77–91. https://doi.org/10.1111/j.1540-6261.1952.tb01525.x doi: 10.1111/j.1540-6261.1952.tb01525.x
    [110] Martens D, Van Gestel T, De Backer M, et al. (2010) Credit rating prediction using Ant Colony Optimisation. J Oper Res Soc 61: 561–573.
    [111] Mazri C (2017) (Re)Defining Emerging Risk. Risk Anal 37: 2053–2065. https://doi.org/10.1111/risa.12759 doi: 10.1111/risa.12759
    [112] McNeil AJ, Frey R, Embrechts P (2015) Quantitative Risk Management (2nd ed): Concepts, Techniques and Tools, Princeton University Press. Princeton.
    [113] Merton RC (1974) On the pricing of corporate debt: the risk structure of interest rate. J Financ 29: 637–654. https://doi.org/10.1111/j.1540-6261.1974.tb03058.x doi: 10.1111/j.1540-6261.1974.tb03058.x
    [114] Mikes A (2005) Enterprise Risk Management in Action. Available from: https://etheses.lse.ac.uk/2924/.
    [115] Mitchell M (2006) Complex systems: Network thinking. Artif Intell 170: 1194–1212. https://doi.org/10.1016/j.artint.2006.10.002 doi: 10.1016/j.artint.2006.10.002
    [116] Mitchell M (2009) Complexity: A Guided Tour. Oxford University Press.
    [117] Mitleton-Kelly E (2003) Complex systems and evolutionary perspectives on organisations: the application of complexity theory to organisations. Oxford: Pergamon.
    [118] Mohajan H (2017) Two Criteria for Good Measurements in Research: Validity and Reliability. Annals of Spiru Haret University 17: 58–82.
    [119] Mohajan HK (2018) Qualitative Research Methodology in Social Sciences and Related Subjects. J Econ Dev Environ Peopl 7: 23–48.
    [120] Morin E (2014) Complex Thinking for a Complex World-About Reductionism, Disjunction and Systemism. Systema 2: 14–22.
    [121] Morin E (1992) From the concept of system to the paradigm of complexity. J Soc Evol Sys 15: 371–385. https://doi.org/10.1016/1061-7361(92)90024-8 doi: 10.1016/1061-7361(92)90024-8
    [122] Motet G, Bieder C (2017) The Illusion of Risk Control, Springer Briefs in Applied Sciences and Technology. Springer, Cham.
    [123] Mousavi S, Gigerenzer G (2014) Risk, Uncertainty and Heuristics. J Bus Res 67: 1671–1678. https://doi.org/10.1016/j.jbusres.2014.02.013 doi: 10.1016/j.jbusres.2014.02.013
    [124] Muth JF (1961) Rational Expectations and the Theory of Price Movements. Econometrica 29: 315–35.
    [125] Newman MJ (2011) Complex systems: A survey. Am J Phy 79: 800–810. https://doi.org/10.1119/1.3590372 doi: 10.1119/1.3590372
    [126] Pagel M, Atkinson Q, Meade A (2007) Frequency of word-use predicts rates of lexical evolution throughout Indo-European history. Nature 449: 717–720.
    [127] Pallant J (2010) SPSS Survival Manual: A Step-by-Step Guide to Data Analysis using SPSS for Windows, Maidenhead: Open University Press.
    [128] Peters E (1991) Fractal Market Analysis: Applying Chaos Theory to Investment and Economics. London: Wiley Finance. John Wiley Science.
    [129] Peters E (1994) Fractal Market Analysis: Applying Chaos Theory to Investment and Economics. New York: John Wiley & Sons.
    [130] Pichler A (2017) A quantitative comparison of risk measures. Ann Oper Res 254: 1–2. https://doi.org/10.1007/s10479-017-2397-3 doi: 10.1007/s10479-017-2397-3
    [131] Plosser M, Santos J (2018) The Cost of Bank Regulatory Capital. Mimeo, Federal Reserve Bank of New York. Staff Report Number 853.
    [132] Ponto J 2015) Understanding and Evaluatin Survey Research. J Adv Pract Oncol 6: 66–171. https://doi.org/10.6004/jadpro.2015.6.2.9 doi: 10.6004/jadpro.2015.6.2.9
    [133] Quinlan C (2011) Business Research Methods. South Westen Cengage Learning.
    [134] Ramsey FP (1926) Truth and Probability. In Ramsey 1931, The Foundations of Mathematics and Other Logical Essays, Chapter Ⅶ, 156–198, edited by R.B. Braithwaite, London: Kegan, Paul, Trench, Trubner and Co., New York: Harcourt, Brace and Company.
    [135] Righi BM, Muller FM, da Silveira VG, et al. (2019) The effect of organisational studies on financial risk measures estimation. Rev Bus Manage 21: 103–117. https://doi.org/10.7819/rbgn.v0i0.3953 doi: 10.7819/rbgn.v0i0.3953
    [136] Ross SA (1976) The arbitrage theory of capital asset pricing. J Econ Theory 13: 341–360. https://doi.org/10.1016/0022-0531(76)90046-6 doi: 10.1016/0022-0531(76)90046-6
    [137] Saunders M, Lewis P, Thornhill A (2019) Research Methods for Business Students, New York Pearson Publishers.
    [138] Saunders M, Lewis P, Thornhill A (2012) Research Methods, Pearson Publishers.
    [139] Savage LJ (1954) The Foundations of Statistics. New York: John Wiley and Sons.
    [140] Sekaran U (2003) Research Methods for Business (4th ed.), Hoboken, NJ: John Wiley &Sons. Shane, S. and Cable.
    [141] Senge P (1990) The Fifth Discipline- The Art and Practice of the Learning Organisation. New York, Currency Doubleday.
    [142] Shackle GLS (1949) Expectations in Economics. Cambridge: Cambridge University Press.
    [143] Shackle GLS (1949) Probability and uncertainty. Metroeconomica 1: 161–173. https://doi.org/10.1111/j.1467-999X.1949.tb00040.x doi: 10.1111/j.1467-999X.1949.tb00040.x
    [144] Sharpe WF (1964) Capital Asset Prices: A Theory of Market Equilibrium under Conditions of Risk. J Financ 19: 425–442. https://doi.org/10.1111/j.1540-6261.1964.tb02865.x doi: 10.1111/j.1540-6261.1964.tb02865.x
    [145] Shefrin H (2010) Behaviouralising Finance. Found Trends Financ 4: 1–184. http://dx.doi.org/10.1561/0500000030 doi: 10.1561/0500000030
    [146] Shefrin H, Statman M (1985) The disposition to sell winners to early and ride loses too long: Theory and evidence. J Financ 40: 777–790. https://doi.org/10.2307/2327802 doi: 10.2307/2327802
    [147] Snowden D, Boone M (2007) A leader's framework for decision making. Harvard Bus Rev 85: 68–76.
    [148] Soramaki K, Cook S (2018) Network Theory and Financial Risk, Riskbooks.
    [149] Sornette D (2003) Why Stock Markets Crash: Critical Events in Complex Financial Systems, Princeton University Press.
    [150] Sornette D, Ouillon G (2012) Dragon-kings: mechanisms, statistical methods, and empirical evidence. Eur Phys J Spec Topics 205: 1–26. https://doi.org/10.1140/epjst/e2012-01559-5 doi: 10.1140/epjst/e2012-01559-5
    [151] Sweeting P (2011) Financial Enterprise Risk Management. Cambridge University Press.
    [152] Tepetepe G, Simenti-Phiri E, Morton D (2021) Survey on Complexity Science Adoption in Emerging Risks Management of Zimbabwean Banks. J Global Bus Technol 17: 41–60
    [153] Thaler RH (2015) Misbehaving: The Making of Behavioural Economics. New York: W. W. Norton and Company.
    [154] Thompson M (2018) How banks and other financial institutions. Brit Actuar J 23: 1–16.
    [155] Tukey JW (1977) Exploratory Data Analysis, MA: Addison-Wesley.
    [156] Turner RJ, Baker MR (2019) Complexity Theory: An Overview with Potential Applications for the Social Sciences. Sys Rev 7: 1–22. https://doi.org/10.3390/systems7010004 doi: 10.3390/systems7010004
    [157] Virigineni M, Rao MB (2017) Contemporary Developments in Behavioural Finance. Int J Econ Financ Issues 7: 448–459.
    [158] Yamane T (1967) Statistics: An Introductory Analysis (2nd ed), New York: Harper and Row.
    [159] Zadeh LA (1965) Fuzzy sets. Inf Control 8: 335–338. http://dx.doi.org/10.1016/S0019-9958(65)90241-X doi: 10.1016/S0019-9958(65)90241-X
    [160] Zadeh LA (1978) Fuzzy sets as basis for a theory of possibility. Fuzzy Set Syst 1: 3–28. https://doi.org/10.1016/0165-0114(78)90029-5 doi: 10.1016/0165-0114(78)90029-5
    [161] Zikmund W (2003) Business Research Methods, 7th Ed, Mason, OH. and Great Britain, Thomson/South-Western.
    [162] Zimbabwe (2011) Technical Guidance on the Implementation of the Revised Adequacy Framework in Zimbabwe. Reserve Bank of Zimbabwe.
    [163] Zimmerman B (1999) Complexity science: A route through hard times and uncertainty. Health Forum J 42: 42–46.
  • This article has been cited by:

    1. Xiaotong Bian, Debao Chen, Feng Zou, Fangzhen Ge, Yuhui Zheng, Fuqiang Liu, Multitask particle swarm optimization algorithm leveraging variable chunking and local meta-knowledge transfer, 2025, 92, 22106502, 101823, 10.1016/j.swevo.2024.101823
  • Reader Comments
  • © 2022 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(3036) PDF downloads(113) Cited by(1)

Figures and Tables

Figures(2)  /  Tables(4)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog