A dynamic multi-objective evolutionary algorithm using a change severity-based adaptive population management strategy

Abstract

In addition to the need for simultaneously optimizing several competing objectives, many real-world problems are also dynamic in nature. These problems are called dynamic multi-objective optimization problems. Applying evolutionary algorithms to solve dynamic optimization problems has obtained great attention among many researchers. However, most of works are restricted to the single-objective case. In this work, we propose an adaptive hybrid population management strategy using memory, local search and random strategies, to effectively handle environment dynamicity for the multi-objective case where objective functions change over time. Moreover, the proposed strategy is based on a new technique that detects the change severity, according to which it adjusts the number of memory and random solutions to be used. This ensures, on the one hand, a high level of convergence and on the other hand, the required diversity. We propose a dynamic version of the Non dominated Sorting Genetic Algorithm II, within which we integrate the above-mentioned strategies. Empirical results show that our proposal based on the use of the adaptive strategy is able to handle dynamic environments and to track the Pareto front as it changes over time. Moreover, when confronted with several recently proposed dynamic algorithms, it has presented competitive and better results on most problems.Keywords Change severity sensing · Dynamic multiobjective optimization · Time-changing objective functions · Adaptive population management · Memory-based strategy · Local search-based strategy · Change severity-based strategy

A Similarity-Based Cooperative Co-Evolutionary Algorithm for Dynamic Interval Multiobjective Optimization Problems

“…The helpful information offered by the archive can assist in handling neighbor sub-problems by cooperation. Azzouz et al proposed an adaptive strategy for managing hybrid populations with memory, local search and random strategies, to effectively tackle DMOPs, which guarantees a rapid convergence and good diversity [45]. Koo et al proposed a selective memory technique, which selects a partial retrieval based on the diversity in the decision space to maintain effective memories [46].…”

Transfer Learning-Based Dynamic Multiobjective Optimization Algorithms

“…slow convergence and poor diversity, when the environment changes. As a result the authors in [1] proposed an adaptive hybrid population management strategy using memory, local search and random strategies to effectively handle environment dynamicity in DMOPs. The special feature of this algorithm is that it can adjust the number of memory and random solutions to be used according to the change severity.…”

Solving Dynamic Multi-objective Optimization Problems Using Incremental Support Vector Machine

“…It has been proved that when the optimal solution returns to the previous position repeatedly or the environment changes periodically, this algorithm will help save computing time and bias search process, thus becoming very efficient. In [15], Azzouz et al proposed an adaptive hybrid population management strategy, which is based on a technology that can measure the severity of environmental changes, then according to the technology it can adjust memory, local search (LS) and the number of random solutions.…”

“…Input: The Dynamic Multi-objective Optimaztion Function F (X); Output: P OSs: the POSs of F (X); 1 Randomly initiate a Population the P op 0 ; 2 P OS 0 =DMOEA(P op 0 ); 3 P OS s =P OS 0 ; 4 Train a SVM classifier SC S by using P g ∈ P OS 0 and N g / ∈ P OS 0 ; 5 Randomly generate solutions {xy 1 , · · · , xy p } of the function F (X) 1 ; 6 if xy i pass the recognition of the SVM SC S then 7 Put xy i into P op 1 8 end 9 P OS 1 =DMOEA(P op 1 ); 10 P OS t =P OS 1 ; 11 for t = 1 to n do 12 P SAM P LES t =P OS t ; 13 Train SC S by using P g ∈ P SAM P LES t and N g ∈ N SAM P LES t ; 14 Randomly generate solutions {xy 1 , · · · , xy p } of the function F (X) t+1 ; 15 if xy i pass the recognition of the SVM SC S then 16 Put xy i into P op t+1 17 end 18 P OS t+1 = DMOEA(P op t+1 ); 19 P OSs = P OSs ∪ P OS t+1 ; 20 end 21 return P OSs;…”