Roulette-tournament Selection for Shrimp Diet Formulation Problem

This paper aims to propose a new selection procedure for real value encoding problem, specifically for shrimp diet problem. This new selection is a hybrid between two well-known selection procedure; roulette wheel selection and binary tournament selection. Shrimp diet problem is investigated to understand the hard constraints and the soft constraints involved. The comparison between other existing selections is also described for evaluation purposes. The result shows that roulette-tournament selection is better in terms of number of feasible solutions achieved and thus suitable for real value encoding problem. However, the combination with other crossover or mutation might be investigated to find the most suited combination that can obtain better best so far solution.