Search

Traditional federated learning faces challenges such as high communication costs, structural heterogeneity,and insufficient privacy protection. To address these issues, this paper proposes a federated learning evolutionary algorithm that applies sparse evolutionary training algorithm to reduce communication costs and integrates local differential privacy protection for participants’ privacy. Additionally, it utilizes the NSGA-Ⅲ algorithm to optimize the network structure and sparsity of the global federated learning model, adjusting the relationship between data availability and privacy protection. This achieves a balance between the effectiveness, communication costs, and privacy of the global federated learning model. Experimental results under unstable communication environments demonstrate that, on the MNIST and CIFAR-10 datasets, compared to the solution with the lowest error rate using the FNSGA-Ⅲ algorithm, the proposed algorithm improves communication efficiency by 57. 19% and 52. 17%, respectively. The participants also achieved（3. 46, 10-4） and（6. 52, 10-4）-local differential privacy. This algorithm can effectively reduce communication costs and protect participant privacy without significantly compromising the accuracy of the global model. [ABSTRACT FROM AUTHOR]

At present, the mainstream solution of edge computing offloading is to model it as a multi-objective optimization problem to minimize energy consumption and delay. Different from the existing research, this paper mainly considered that tasks in different unloading areas had certain similarity in edge computing, which could be used to accelerate the convergence speed and solution effect of the algorithm. Based on this, this paper proposed an evolutionary multi-task multi-objective optimization algorithm to solve task offloading problems in different regions based on evolutionary multi-task optimization. This algorithm considered multiple independent regions to be optimized. It modeled the task offloading system model for each region as a multi-objective optimization problem, dynamically adjusting the communication level of the population by learning the user distribution in different regions and the similarity of the tasks to be processed, accelerating convergence speed, and achieving optimization for two different regions through one evolution. The experimental results show that the proposed algorithm has achieved good results in convergence speed and the uniformity of the optimal solution distribution, and can obtain the unloading deployment optimization scheme under edge computing. [ABSTRACT FROM AUTHOR]

To address the low-quality problem of automatic extraction of ABAC policies, this paper proposed an automatic ex traction and optimization enhancement method of ABAC policies driven by access control logs. Firstly, the method constructed an ensemble learning model to map user behavior and permission allocation into a policy logic tree, identifying the relevance and underlying patterns of access authorization decisions to generate preliminary policies. Secondly, the method employed two algorithms of single attribute optimization and rule hinary reduction to deeply optimize the policies, simplifying policy structure and compressing policy scale. Finally, the method proposed an error metric-based rule conflict resolution approach to enhance mutually exclusive, complete ABAC policies. Furthermore, the method used policy performance balancing algorithm of a multi-objective optimization to achieve the optimal model selection for different scenario requirements. This paper tested and verified on balanced datasets and sparse datasets respectively, the experimental results demonstrate that the method achieves a highest accuracy of 96.69% on the balanced dataset, compresses the policy seale to 19.7% of the original. On the sparse dataset, the highest accuracy reaches 87.74%, with the policy scale compresses to 23% of the original. This method balances the prediction accuracy and structural simplicity of policies, and is applicable for both balanced logs and sparse logs, ensuring the access control system can implement efficient and secure access authorization management in practical applications. [ABSTRACT FROM AUTHOR]

Most of the existing MCS task allocation methods often only consider the unilateral utility of workers or platforms, and the composition of utility is not comprehensive enough. Therefore, this paper designed a heterogeneous utility mechanism for both workers and platforms based on the worker reputation index and task proficiency index. It proposed a dual-population competitive multi-objective evolutionary algorithm (DCMEA) to obtain the optimal worker and platform heterogeneous utilities. The algorithm firstly initialized the population using a stochastic greedy algorithm, then divided the population into a winner population and a loser population using a binary bidding tournament algorithm, and employed different evolutionary strategies for each population. Finally, this paper proposed the repair operator to make the invalid individuals in the evolution process satisfy the constraint, Experiments on real-world datasets show that DCMEA converges faster than the baseline algorithm, and can find more accurate and stable task allocation solution sets, while maintaining its performance in more complex scenarios. [ABSTRACT FROM AUTHOR]

In consideration of the superior performance of the wolf pack algorithm on single-objective optimization problems, combining with the biological habits of wolves to apply it to multi-objective optimization problems, this paper proposed a multiobjective wolf pack algorithm with elite guidance and information interaction (MOWPA-EGII). Firstly, MOWPA-EGII proposed an elite guidance strategy, using the elite wolf in the external file and the head wolf of the current sub-population to jointly guide the population movement, so that the artificial wolves uniformly distributed in the whole search space, and the strategy enhanced the global search ability of the algorithm; Secondly, it designed the information interaction mechanism to simulate the information transfer in the wolf hunting process, so that individuals with different advantages could transfer information to each other to ensure the hunting efficiency of the wolves and improved the ability of the algorithm to explore the optimal solution of the Pareto; Finally, it added the mutation operator to perturb the moving direction of the artificial wolves, so as to let the algorithm jump out of the local optimum and enhance the local search ability of the algorithm. In order to verify the effectiveness of MOWPA-EGII, comparing it with 5 classical algorithms and 10 recent algorithms, and the results show that MOWPA-EGII possesses good convergence and diversity, which proves that the present algorithm has a better optimization performance. [ABSTRACT FROM AUTHOR]

To solve the parameter tunning problem of PID controllers in master-slave control structure of electro-hydraulic servo system, this paper proposed a solution method based on multi-objective optimization algorithm combined with model simulation. This method modeled the problem as a multi-objective optimization problem whose target space included four dimensions, namely master-slave synchronization error, settling time, overshoot and integral squared error. And it established the simulation model of master-slave PID control structure to obtain the target value of the target space. Additionally, it improved the MOEAD algorithm by strengthenning the diversity of generated offspring through the selection of parents along with the offspring generate operators and readjusting the weight vector through the substitution strategy of invalid ones using external populations and binding mechanism between weight vectors and individuals. Finally, this method got the final Pareto non-dominated solution set and then selected the knee point individual as the optimal parameter of the controller in the solution set, which improves the control quality of the master-slave synchronization structure. [ABSTRACT FROM AUTHOR]

To address the issue of resource wastage and increasing latencies in the placement process of edge servers, this paper analyzed and modeled the user density and average access time of edge server placement schemes as multi-objective optimization problems. It designed an edge server placement scheme based on user density and average access time, and proposed a multi-objective sea-horse genetic algorithm (MOSGA) to solve this problem. The MOSGA algorithm firstly used the idea of multi-objective optimization algorithm to improve the SHO algorithm, so that the SHO algorithm could be applied to multi-objective optimization problems. It used the genetic algorithm to improve the propagation operation of the SHO algorithm, so that the MOSGA could better jump out of the local optimal solution and accelerate the solution of the problem. The proposed algorithm was verified on the data set of Shanghai Telecom, and the simulation experiment results show that MOSGA is obviously better than RA, K-means, NSGA and LMM, which not only effectively solves the problem of server resource waste, but also greatly reduces the time of terminal equipment to access the server. [ABSTRACT FROM AUTHOR]

To optimize the distributed flexible job shop scheduling problem considering both the makespan and machine energy consumption, this paper proposed an improved multi-objective Squirrel Search Algorithm. The algorithm introduced a conversion mechanism based on ranked order value to achieve the conversion between squirrel position vectors and scheduling solutions, and designed decoding strategy from semi-active to active for machine idle time.Additionally, it devised three population initialization strategies based on different optimization objectives. The proposed dynamic predator strategy can better coordinate the algorithm’s global exploration and local exploitation capabilities. Finally, the algorithm used four local searches to increase population diversity. Experimental results on 20instances show that the improved algorithm achieves better quality and diversity of solutions, proving that it can effectively solve the distributed energy-efficient flexible scheduling problem. [ABSTRACT FROM AUTHOR]