An evolutionary multitasking algorithm based on k-nearest neighbors pre-selection strategy for constrained multi-objective optimization

Expert Systems with Applications, Journal Year: 2025, Volume and Issue: unknown, P. 127768 - 127768

Published: April 1, 2025

Language: Английский

---

Adaptive Social Mobility-Restructuring Differential Evolution

Published: Jan. 1, 2025

Language: Английский

---

Comparative Study of Modern Differential Evolution Algorithms: Perspectives on Mechanisms and Performance

Mathematics, Journal Year: 2025, Volume and Issue: 13(10), P. 1556 - 1556

Published: May 9, 2025

Since the discovery of Differential Evolution algorithm, new and improved versions have continuously emerged. In this paper, we review selected algorithms based on that been proposed in recent years. We examine mechanisms integrated into them compare performance algorithms. To their performances, statistical comparisons were used as they enable us to draw reliable conclusions about algorithms’ performances. use Wilcoxon signed-rank test for pairwise Friedman multiple comparisons. Subsequently, Mann–Whitney U-score was added. conducted not only a cumulative analysis algorithms, but also focused performances regarding function family (i.e., unimodal, multimodal, hybrid, composition functions). Experimental results obtained problems defined CEC’24 Special Session Competition Single Objective Real Parameter Numerical Optimization. Problem dimensions 10, 30, 50, 100 analyzed. highlight promising further development improvements study

Language: Английский

---

A novel metaheuristic population algorithm for optimising the connection weights of neural networks

Evolving Systems, Journal Year: 2024, Volume and Issue: 16(1)

Published: Dec. 5, 2024

Abstract The efficacy of feed-forward multi-layer neural networks relies heavily on their training procedure, where identifying appropriate weights and biases plays a pivotal role. Nonetheless, conventional algorithms such as backpropagation encounter limitations, including getting trapped in sub-optimal solutions. To rectify these inadequacies, metaheuristic population are advocated dependable alternative. In this paper, we introduce novel methodology termed, DDE-OP, which leverages the principles differential evolution enriched with division-based scheme an opposite-direction strategy. Our approach integrates two effective concepts evolution. Initially, proposed algorithm identifies partitions within search space through clustering designates obtained cluster centres to serve representatives. Subsequently, updating incorporates clusters into current population. Lastly, quasi-opposite-direction strategy is used augment exploration. Extensive evaluation diverse classification approximation tasks demonstrate that DDE-OP surpasses population-based methodologies.

Language: Английский

---

Forest Fire Ash Optimizer (FFA): A Novel Physics-based Metaheuristic Algorithm for Implementing Exploration-Exploitation Flexible Regulation

Research Square (Research Square), Journal Year: 2024, Volume and Issue: unknown

Published: Sept. 30, 2024

This study proposes a novel physics-inspired metaheuristic algorithm named Forest Fire Ash Optimizer (FFA). Inspired by the characteristics of ash movement in different forest fire burning stages, intelligently transformed these regular behaviors into variety unique algorithmic mechanisms, including four-population mechanism based on fitness and denseness partitioning, an inverse proportionality selection for elite influence, strategy grounded actual physical phenomena. These components complement each other to enable flexible regulation exploration exploitation, i.e., two phases are not merely sequential, but change dynamically depending search status FFA agents. To verify effectiveness proposed algorithm, is qualitatively analyzed using CEC-2022 test suite. Additionally, classical standard suite, CEC-2017 suite conducted compare performance with 9 advanced algorithms. The results demonstrate that excels performance, exhibiting high stability, flexibility, robustness. Finally, applied challenging real-world engineering optimization problems. indicate that, compared competing algorithms, provides superior more solutions, predicting its potential in-depth applications fields.

Language: Английский

---

An Experimental Comparison of Self-Adaptive Differential Evolution Algorithms to Induce Oblique Decision Trees

Mathematical and Computational Applications, Journal Year: 2024, Volume and Issue: 29(6), P. 103 - 103

Published: Nov. 9, 2024

This study addresses the challenge of generating accurate and compact oblique decision trees using self-adaptive differential evolution algorithms. Although traditional tree induction methods create explainable models, they often fail to achieve optimal classification accuracy. To overcome these limitations, other strategies, such as those based on evolutionary computation, have been proposed in literature. In particular, we evaluate use variants evolve a population encoded real-valued vectors. Our proposal includes (1) an alternative initialization strategy that reduces redundant nodes (2) fitness function penalizes excessive leaf nodes, promoting smaller more trees. We perform comparative performance analysis variants, showing while exhibit similar statistical behavior, Single-Objective real-parameter optimization (jSO) method produces most is second best compactness. The findings highlight potential algorithms improve effectiveness machine learning applications.

Language: Английский

---

Improving the selection of differential evolution through a quartile-based ranked operator

International Journal of Machine Learning and Cybernetics, Journal Year: 2024, Volume and Issue: unknown

Published: Nov. 30, 2024

Language: Английский

---