Tutorial Sobre o Uso de Técnicas para Controle de Parâmetros em Algoritmos de Inteligência de Enxame e Computação Evolutiva

Abstract

Nature has always been a great source of inspiration for the development of computational approaches for optimization. Two major groups representing this class of biologically inspired algorithms are Swarm Intelligence and Evolutionary Computation. Such algorithms are called metaheuristics and are recognized to be efficient approaches for solving complex problems. Both Swarm Intelligence and Evolutionary Computation share common features such as the use of stochastic components during the optimization process and various parameters for configuration. The setup of parameters of an algorithm has an important role in defining its behavior, guiding the search and biasing the quality of the solutions found. However, adjusting the parameters is not a simple task, becoming an optimization problem within the problem being optimized. In addition, an appropriate setting for the parameters may change during the optimization process making this task even harder. There are two ways to adjust the parameters of an algorithm. The offline control that is performed before running the algorithm and parameter values remains fixed and the online control where parameter values may change during the optimization process. This article focuses on reviewing the online parameter control strategies applied in Evolutionary Computation and Swarm Intelligence. As a result, this review analyzes and points out the key techniques and algorithms used and suggests some directions for future research.

Data Prediction in Datasets of Internet of Things with Recurrent Neural Networks

“…The parameter related to the number of iterations for the training set was implemented in an online and deterministic way, based on [41]. The algorithm measures the mean absolute error and concludes the training set if the error doesn't decrease after a certain number of iterations.…”

The Multiple Knapsack Problem Approached by a Binary Differential Evolution Algorithm with Adaptive Parameters

“…The equality ensures that at least one dimension will be perturbed. The perturbation is carried out by the bit-flip mutation using its probability (line [11][12] or by the crossover operator (line 14). if (random(0, 100) < CR ) {Uniform Crossover} then 9:…”

“…In the on-line control, or parameter control, the values for the parameters change throughout the execution of the algorithm. The control of parameters during the optimization process has been consistently used by several optimization algorithms and applied in different problem domains [12], [13], [14], [15], [16]. In this way, a method to adapt the control parameters (crossover and mutation rates) of DE is applied.…”