Search

Comparative view of the cost function when using the GWO, GD, and PSO algorithms. [Display omitted] • A Policy Iteration Reinforcement Learning-based control approach is proposed. • A Grey Wolf Optimizer algorithm trains the policy neural network. • A comparison with classical Gradient Descent and metaheuristic PSO is offered. • The validation is done on real-time nonlinear servo system position control. • The fast convergence of the suggested approach is proved. This paper presents a new Reinforcement Learning (RL)-based control approach that uses the Policy Iteration (PI) and a metaheuristic Grey Wolf Optimizer (GWO) algorithm to train the Neural Networks (NNs). Due to an efficient tradeoff to exploration and exploitation, the GWO algorithm shows good results in NN training and solving complex optimization problems. The proposed approach is compared to the classical PI RL-based control approach using the Gradient Descent (GD) algorithm, and with the RL-based control approach which uses the metaheuristic Particle Swarm Optimization (PSO) algorithm. The experiments are conducted using a nonlinear servo system laboratory equipment. Each approach evaluated on how well it solves the optimal reference tracking control for an experimental servo system position control system. The policy NNs specific to all three approaches are implemented as state feedback with integrator controllers to remove the steady-state control errors and thus ensure the convergence of the objective function. Because of the random nature of metaheuristic algorithms, the experiments for GWO and PSO algorithms are run multiple times and the results are averaged before the conclusions are presented. The experimental results shows that for the control objective considered in this paper, the GWO algorithm represents a better solution compared to GD and PSO algorithms. [ABSTRACT FROM AUTHOR]

Following two decades of sustained studies, metaheuristic algorithms have made considerable achievements in the field of multi-objective optimization problems (MOPs). However, under most existing metaheuristic frameworks, an improved scheme introduced to address specific defects usually leads to additional problems that need to be solved further. Emerging optimization mechanisms should be considered to break the bottleneck, and an adaptive multi-objective competitive swarm optimization (AMOCSO) algorithm, a promising option for solving MOPs, is proposed in this paper. Firstly, the competitive mechanism is modified so that it can perform well on MOPs, and an improved learning scheme is designed for the winners and the losers, which can greatly enhance the optimization efficiency and balance the convergence and the diversity of the proposed algorithm. Then, an external archive and its maintenance schemes are introduced to prevent the population from degenerating and make the algorithm framework more comprehensive. Moreover, a practical adaptive strategy is proposed to fill the blank of parameter research, and no human factors exist in AMOCSO, which means that an amazing promotion can be achieved in generalization. Finally, abundant experimental studies are carried out, and the results of comparative experiments show that the proposed algorithm has significant advantages over several state-of-the-art algorithms. [ABSTRACT FROM AUTHOR]

• A combination of Deep Q-Learning algorithm and metaheuristic GSA is offered. • GSA initializes the weights and the biases of the neural networks. • A comparison with classical random, metaheuristic PSO and GWO is carried out. • The validation is done on real-time nonlinear servo system position control. • The drawbacks of randomly initialized neural networks are mitigated. This paper presents a novel Reinforcement Learning (RL)-based control approach that uses a combination of a Deep Q-Learning (DQL) algorithm and a metaheuristic Gravitational Search Algorithm (GSA). The GSA is employed to initialize the weights and the biases of the Neural Network (NN) involved in DQL in order to avoid the instability, which is the main drawback of the traditional randomly initialized NNs. The quality of a particular set of weights and biases is measured at each iteration of the GSA-based initialization using a fitness function aiming to achieve the predefined optimal control or learning objective. The data generated during the RL process is used in training a NN-based controller that will be able to autonomously achieve the optimal reference tracking control objective. The proposed approach is compared with other similar techniques which use different algorithms in the initialization step, namely the traditional random algorithm, the Grey Wolf Optimizer algorithm, and the Particle Swarm Optimization algorithm. The NN-based controllers based on each of these techniques are compared using performance indices specific to optimal control as settling time, rise time, peak time, overshoot, and minimum cost function value. Real-time experiments are conducted in order to validate and test the proposed new approach in the framework of the optimal reference tracking control of a nonlinear position servo system. The experimental results show the superiority of this approach versus the other three competing approaches. [ABSTRACT FROM AUTHOR]

• A new algorithm, called BTLBO, is proposed to balance between exploration and exploitation. • A new Tutoring Phase is added to TLBO to accelerate the convergence to the global optimum. • A new Restarting Phase is added to TLBO to promote convergence to the global optimum. • Control parameters are self-adaptively tuned. • The superiority of BTLBO is verified by solving standard benchmarks and real world optimization problems. Teaching-learning-based optimization (TLBO) is a population-based metaheuristic algorithm which simulates the teaching and learning mechanisms in a classroom. The TLBO algorithm has emerged as one of the most efficient and attractive optimization techniques. Even though the TLBO algorithm has an acceptable exploration capability and fast convergence speed, there may be a possibility to converge into a local optimum during solving complex optimization problems and there is a need to keep a balance between exploration and exploitation capabilities. Hence, a Balanced Teaching-Learning-Based Optimization (BTLBO) algorithm is proposed in this paper. The proposed BTLBO algorithm is a modification of the TLBO algorithm and it consists of four phases: (1) Teacher Phase in which a weighted mean is used instead of a mean value for keeping the diversity, (2) Learner Phase, which is same as the learner phase of basic TLBO algorithm, (3) Tutoring Phase, which is a powerful local search for exploiting the regions around the best ever found solution, and (4) Restarting Phase, which improves exploration capability by replacing inactive learners with new randomly initialized learners. An acceptable balance between the exploration and exploitation capabilities is achieved by the proposed BTLBO algorithm. To evaluate the performance of BTLBO algorithm, several experimental studies are conducted on standard benchmark suits and the results are compared with several TLBO variants and state-of-the-art population-based optimization algorithms. The results are in excellent agreement and confirm the efficiency of BTLBO algorithm with accelerated exploitation and exploration capabilities with an appropriate balance between such criteria for solving complex optimization problems. [ABSTRACT FROM AUTHOR]

This paper presents a new fractal search space decomposition-based algorithm to address the issue of scaling up the divide and conquer approach to deal with large scale problems (up to 50 continuous decision variables). The proposed algorithm, called polyFrac, fractally decomposes the search space using hyper-polytopes. It allows moving throughout different granularity levels by only computing the average of vertices of a hyper-polytope to obtain the coordinates of the centroids. Only the most promising hyper-polytopes are decomposed into child-polytopes. Then, a simple deterministic local search (single solution-based metaheuristic) is used to perform the intensification process to find the best solution within the selected lowest hyper-polytope. The proposed algorithm performance is evaluated on the well-known SOCO 2011, CEC 2013, and CEC 2017 benchmarks and compared with 26 states of the art algorithms. A real-world optimization problem is also used to calibrate its performance. The obtained results show that polyFrac outperforms all the algorithms. Moreover, experimental results and analysis suggest that polyFrac is a highly competitive optimization algorithm for solving large-scale and complex optimization problems. [ABSTRACT FROM AUTHOR]

For real parameter single objective optimization, Differential Evolution (DE) performs better than other types of population-based metaheuristic. Nevertheless, in the field of DE, it is impossible for a given combination of operators to perform well in all fitness landscapes. Practice has proved that assembling more than one combinations of operators may lead to improvement in solution. However, some DE algorithms with good performance, e.g., one of the best performers in competitions of real parameter single objective optimization among population-based metaheuristics held by the series of IEEE Congress on Evolutionary Computation - L-SHADE-EpSin, are still not involved in ensemble. Furthermore, ensemble based on similar DE algorithms is rarely reported. Based on such a background, we propose L-SHADE-E, ensemble of L-SHADE-EpSin and L-SHADE-RSP in this paper. The two constituent algorithms are both variants of L-SHADE and belong to similar DE algorithms. In our algorithm, L-SHADE-EpSin or L-SHADE-RSP is selected randomly to run at the beginning. If progress on fitness is low for generations, the other constituent algorithm takes over population immediately. Our experiments are based on the CEC 2014 and CEC 2017 benchmark test suites. Altogether, our algorithm is compared with nine DE algorithms and three population-based metaheuristics other than DE. Experimental results show that our algorithm is very competitive. [ABSTRACT FROM AUTHOR]

Differential privacy (DP), as a promising privacy-preserving model, has attracted great interest from researchers in recent years. At present, research on the combination of deep learning and DP is active. In contrast, another widely used artificial intelligence technique, swarm intelligence (SI), has received little attention in the context of DP even though it also triggers privacy concerns. For this reason, this paper attempts to combine DP and SI for the first time and proposes a general differentially private swarm intelligence algorithm framework (DPSIAF). By utilizing the exponential mechanism, this framework can easily develop existing SI algorithms into the private versions. As examples, we apply it to four popular SI algorithms, and the related analyses demonstrate the effectiveness of our DPSIAF. More importantly, our private algorithms also exhibit some interesting experimental phenomena. In some cases, their performance is not strictly affected by the privacy budget, and one of them even outperforms its non-private version. These findings are different from the conventional cognition, which indicates DP owns the potential to serve as an optimization tool for some SI algorithms. Our study may provide a new perspective on DP and promote the synergy between metaheuristic optimization community and privacy computing community. [ABSTRACT FROM AUTHOR]