Project Title: Return on Investment as Genetic Algorithm Performance Criterion 
Student: Tomas Kulvietis
Course: BSc in Computer Science with Year in Industry

Abstract: 
The aim of this project is to design, implement and analyse a cost-conscious genetic algorithm.
The desirable end result would be an algorithm which would consistently and reliably generate a
candidate solution with matching or higher fitness score in fewer generations than the variant of
the algorithm that doesn't use return of investment (ROI = Investment??Cost
Cost ) as a performance criterion.
Key objectives required to meet the outlined aim:
1. Identify and assess effectiveness of performance optimization approaches currently in use
within context of evolutionary algorithms.
2. Design and implement an open an extensible GA that follows the conventional EA structure
(i.e. generating populations, breeding best-fit individuals within the population, evaluating
candidate fitness).
3. Design and implement second, cost-conscious GA which will build on top of GA developed
in objective 2.
4. Run series of performance tests on the same problems, using matching hardware and starting
parameters using both the conventional GA (objective 2) and EA that utilises ROI as a
performance criterion (objective 3), recording the results obtained from the tests.
5. Summarise and analyse findings made in order to assess the usefulness of ROI as a performance
criterion for GAs.