«Back to research topics list

Evolutionary Optimization Algorithms

Using an outward selective pressure for improving the search quality of the MOEA/D algorithm
A paper published in the Computational Optimization and Applications journal.

Highlights of the paper:
  • A follow-up of the observations made in the previous paper in the Applied Soft Computing journal concerning the influence of specimen neighbourhood definition on the convergence of the MOEA/D algorithm
  • Two techniques used for increasing the pressure towards the edges of the Pareto front
  • Weight vectors were modified in order to include vectors diverging to the outside of the Pareto front
  • Also, neighbourhoods of specimens placed near the edges of the Pareto front were modified
  • The tested methods allowed increasing the hypervolume enclosed by the attained solutions
Modified neigborhoods around specimens placed at the edges of the Pareto front.



The effects of asymmetric neighborhood assignment in the MOEA/D algorithm
A paper published in the Applied Soft Computing journal.

Highlights of the paper:
  • The influence of specimen neighbourhoods on the convergence of the MOEA/D algorithm is studied
  • Theoretical analysis is provided which describes asymmetries in the convergence of the population
  • It is observed that for typically used parameters the asymmetries appear
  • A symmetrical test problem is proposed to study the effects of various parameterizations, with possible influences from the problem itself removed from the experimental setup
  • The observed effects are verified to be statistically significant
  • Guidelines for removing the asymmetries in practical applications are provided
Indication of asymmetries observed for the neighbourhood size T = 10.



Sim-EA: An Evolutionary Algorithm Based on Problem Similarity
The research presented in a paper at the IDEAL 2014 Conference.

Highlights of the paper:
  • A multipopulation algorithm Sim-EA is proposed
  • The algorithm is used for solving several similar instances of an optimization problem
  • A migration mechanism based on the similarity of problem instances is used
  • The proposed migration mechanism is observed to improve the results obtained by the algorithm

An overview of elements of the Sim-EA algorithm



Continuous Population-Based Incremental Learning with Mixture Probability Modeling for Dynamic Optimization Problems
The research presented in a paper at the IDEAL 2014 Conference.

Highlights of the paper:
  • An extension of the PBILC algorithm suitable for multimodal problems is proposed
  • A mixture of Gaussians is used as the probabilistic model
  • A migration mechanism based on the similarity of problem instances is used
  • The proposed migration mechanism is observed to improve the results obtained by the algorithm
«Back to research topics list