China
Africa
Explore chapters and articles related to this topic
Harmony Search
Published in Nazmul Siddique, Hojjat Adeli, Nature-Inspired Computing, 2017
In order to improve the convergence and accuracy of the GHS algorithm, Cobos et al. (2011) proposed a modification to the GHS algorithm using the concepts of a learnable evolution model (LEM). The LEM can locate promising areas where the global optimum is to be found and work with discrete and continuous variables (Michalski, 2000). The LEM is used to determine which individuals in a population or a set of individuals from a previous population are better than others. Thus, the LEM helps in generating new populations. The approach is called GHS+LEM. A set of conjunctive rules is used, which delineate the regions about which there is a greater chance of finding a better value for each xi. Rules are selected using the parameter RCR (rule consideration rate). The procedure of the improvisation step of the GHS+LEM algorithm is presented below: If rand(0,1) < HMCR, then { xi′=xkj, j ∈ {1, …, HMS} and for all i = 1, …, N If rand(0,1) ≤ PAR(t), then xi′=xkbest where bestis the index of the best harmony in HM and k = 1, …, n } Else { If rand(0,1) ≤ RCR, then xi′=(Libest,Uibest) where best is the best set of rules Else x′i = Li + rand(0,1) × (Ui − Li) {random generation from the range} }
Multi-objective mobile robot path planning problem through learnable evolution model
Published in Journal of Experimental & Theoretical Artificial Intelligence, 2019
Learnable evolution model (Michalski, 2000) is a non-Darwinian-type EC algorithm where the evolutionary process is governed through a machine learning method. LEM seeks to determine why certain individuals in a population have better fitness in relation to others. The revealed reasons are expressed as inductive hypotheses and then used in generating new populations. Indeed, LEM avoids semi-blind search and uses the experiences of the past generations in order to generate new populations. An intelligent evolution is conducted through LEM, leading to detection of the right directions for evolution, hence making large improvements in the individuals’ fitness.
Hybrid evolutionary optimisation with learning for production scheduling: state-of-the-art survey on algorithms and applications
Published in International Journal of Production Research, 2018
Learnable Evolution Model (LEM) uses a machine learning to infer rules derived from gene values of the least and most fit individuals and then exploits that knowledge when creating children (Michalski 2000). The machine learning then infers rules that describe allele ranges in some genes that belong to fitter individuals. Gene values in newly created individuals are stochastically generated within those ranges.