Computational Intelligence for Optimization (Group U)
We were asked to apply our Genetic Algorithm (GA) knowledge to solve an optimization problem. Using GAs to evolve solutions to a Sudoku problem: How to represent such a problem in our code, and how to design a fitness function and appropriate genetic operators.
We can conclude that the algorithm showed good results, producing game solutions for different levels of difficulty. It is worth mentioning that tweaking crossover and mutation parameters - including elitism, increasing population size and number of generations, can improve the outcome, at the cost of speed and time, which can be suboptimal.
Additionally, most of the algorithms were getting stuck at local minimums, and further strategies need to be considered in order to overcome this issue and achieve the global minimum sooner, with less iterations.