Academic project focused on the implementation and analysis of a parameterizable evolutionary algorithm. Includes a graphical interface, experimentation features, and a written technical report.
This project explores the application of evolutionary algorithms to solve a target-matching problem. The algorithm iteratively improves a population of candidate solutions using biologically inspired operations such as selection, mutation, and recombination. The main objective is to evaluate the effect of parameter tuning on performance and convergence.
The algorithm is based on representing each individual as a character string. Given a target string, the algorithm evolves an initial random population through generations, aiming to produce an exact match.
Users can configure the following:
- Target string
- Population size
- Mutation rate
- Mutation strategy (replacement or swap)
- Selection method (elitist or probabilistic)
- Crossover method (one-point or two-point)
The interface allows users to:
- Launch experiments with custom settings
- Observe real-time updates of best individuals and population metrics
- Monitor convergence behavior and generation count
- Modular Java implementation using standard object-oriented patterns.
- GUI built with Swing for parameter input and result display.
- Fitness function defined as the number of matching characters with the target string.
- Two mutation operators: character replacement and position swapping.
- Selection based on either top-performers or probabilistic selection proportional to fitness.
- Recombination via one-point or two-point crossover.
The project includes an in-depth analysis of how different parameter combinations affect convergence speed and final accuracy. Key findings include:
- Higher mutation rates can speed up exploration but may slow down convergence after early improvements.
- Elitist selection tends to converge faster but risks premature stagnation.
- Two-point crossover produces slightly better convergence than one-point in complex targets.
Detailed experimental results, tables, and conclusions can be found in the reports.
- Java
- Swing (Java GUI toolkit)
- Eclipse (project structure and build)
The reports/ folder contains:
- Final technical report (PDF)
- Summary document with key tables and analysis