Differential Evolution with an Evolution Path: A DEEP Evolutionary Algorithm.

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Evolutionary Computation

Background:

• Differential Evolution (DE) employs a distributed model (DM) for exploration.
• Evolution Strategy (ES) uses a centralized model (CM) for convergence.
• Existing algorithms often lack a balance between exploration and convergence.

Purpose of the Study:

• To propose a new evolutionary algorithm (EA) framework, DEEP (DE with an EP).
• To enhance DE performance by incorporating cumulative correlation information from ES.
• To develop a self-adapting mechanism for evolutionary algorithms.

Main Methods:

• Introduced the DEEP framework, integrating the Evolution Path (EP) concept from CM-ES into DE.
• Developed a novel EA framework that combines DM and CM advantages without direct hybridization.
• Implemented a self-adaptation mechanism inherent to the DEEP architecture.

Main Results:

• DEEP algorithms demonstrated enhanced performance compared to original DEs.
• Experiments on CEC'13 test suites and practical problems showed promising results.
• The proposed framework offers advantages of both distributed and centralized models.

Conclusions:

• DEEP provides a robust framework for improving evolutionary algorithms.
• The inherent self-adaptation mechanism simplifies parameter tuning.
• DEEP represents a significant advancement in evolutionary computation for optimization tasks.