search

tanzairatier / jmoo2

JMOO = Joe's Multi Objective Optimization. Provides a python-based library of usable problems, algorithms, and tools for composition of valuable MOO-research.
1 stars 0 forks source link

[Algorithm] Get information and add older algorithms #12

Open tanzairatier opened 7 years ago

tanzairatier commented 7 years ago

NSGA SPEA VEGA MOGA FFGA NPGA HLGA SOEA PAES

tanzairatier commented 7 years ago

VEGA = Vector Evaluated Genetic Algorithm

  1. initial population
  2. repeat until stopping criteria: 2a. crossover + mutate to reproduce (duplicate population) 2b. selection by "VESelection": split the population evenly into M subpopulations, where M corresponds to the number of objectives. In each subpopulation, you only consider the i_th objective of its fitness score when running a random tournament.
tanzairatier commented 7 years ago

HLGA = Hajela and Lin's Genetic Algorithm

Every individual gets its own set of weights to each objective. Its fitness is computed as the product sum of those weights*objectives.

HLGA ensures that the weights of each individual be diverse. But not sure how that occurs. There are prior algorithms that simply generate random weights over each individual.