Internal lattice reconfiguration for diversity tuning in Cellular Genetic Algorithms

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Evolutionary Computation

Background:

• Cellular Genetic Algorithms (cGAs) offer high performance, easy implementation, and massive parallelism.
• Balancing exploration and exploitation is crucial for effective evolutionary optimization.
• cGAs possess inherent parameters to maintain diversity during evolution.

Purpose of the Study:

• To propose internal lattice reconfiguration for controlling the exploration-exploitation trade-off in cGAs.
• To assess performance improvements resulting solely from topological changes via reconfiguration.
• To evaluate the influence of reconfiguration using local selection methods with opposing pressures.

Main Methods:

• Implementing internal lattice reconfiguration to manage the exploration-exploitation balance.
• Utilizing simplified genetic operators to isolate the impact of reconfiguration mechanisms.
• Applying two local selection methods with contrasting selection pressures.
• Testing the proposed techniques on continuous, real-world, and combinatorial problems.

Main Results:

• Internal reconfiguration of the lattice effectively controls the exploration-exploitation trade-off.
• Algorithmic performance improvements are attributed to topological changes induced by reconfiguration.
• The proposed methods demonstrate statistical significance in efficiency and efficacy across various problem domains.

Conclusions:

• Internal lattice reconfiguration is a viable strategy for enhancing cGA performance.
• Topological adjustments through reconfiguration are key drivers of optimization improvements.
• The approach is robust, showing effectiveness on a wide range of computational problems.