Pattern generation for multi-class LAD using iterative genetic algorithm with flexible chromosomes and multiple populations

Hierarchical multi-class LAD based on OvA-binary tree using genetic algorithm

“…In particular, we adopt and modify the iterative GA using flexible chromosomes and multiple populations (IGA-FCMP) proposed by Kim and Choi (2015). Details of the GA setting are as follows.…”

“…OvA-type MC-LAD algorithms were suggested by Herrera and Subasi (2013) and Kim and Choi (2015). Herrera and Subasi (2013) proposed an algorithmic approach using MILP to efficiently generate OvA-type LAD patterns in a multi-class dataset.…”

“…However, the MILP-based approach has the disadvantage that it may require excessive computational time and effort to develop the MILP model. Kim and Choi (2015) proposed an iterative genetic algorithm (GA) with flexible chromosomes and multiple populations (IGA-FCMP) to improve the shortcomings of the existing LAD pattern generation methods including the MILP-based approach.…”

“…According to the decomposition type, they can be categorized into OvO-type and OvA-type MC-LAD. Research on OvO-type MC-LAD was conducted by Moreira (2000), Mortada et al (2014), and Kim and Choi (2015). Moreira (2000) proposed two methods to decompose one MCCP into BCPs.…”

“…Herrera and Subasi (2013) proposed an algorithmic approach using MILP to efficiently generate OvA-type LAD patterns in a multi-class dataset. Kim and Choi (2015) applied the IGA-FCMP to generate OvA-type patterns for MC-LAD. Through a numerical experiment, they illustrated that the accuracy of this approach was inferior to that of OvO-type MC-LAD using IGA-FCMP.…”