12.07.2015 Views

Davide Cherubini - PhD Thesis - UniCA Eprints

Davide Cherubini - PhD Thesis - UniCA Eprints

Davide Cherubini - PhD Thesis - UniCA Eprints

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

4.1 Heuristic Algorithmslocal minimum. Each encountered configuration is stored in one or more tabulists, forming the so-called Tabu Search Memory.The Tabu search method does not choose the next step (or move) randomly,but following a specific strategy.4.1.2 Genetic Algorithm (GA)“Genetic algorithms are based on a biological metaphor: they view learning as acompetition among a population of evolving candidate problem solutions. A ’fitness’function evaluates each solution to decide whether it will contribute to thenext generation of solutions. Then, through operations analogous to gene transferin sexual reproduction, the algorithm creates a new population of candidatesolutions.” [56].From this definition, typical genetic algorithm requires two factors to be defined:1. a “genetic” representation of the solution domain,2. a “fitness function” to evaluate the solution domain.A standard genetic representation of the solution is as an array (bits, integer,real, or other types work essentially in the same way) because this data structure,especially if it has fixed length, facilitates the crossover operation. The fitnessfunction is always problem-dependent, it is directly defined over the genetic representation,and it measures the quality of the solution.The main steps followed by the Genetic Algorithm are:1. InitializationInitially many individual solutions are randomly generated to form an initialpopulation over the search space (solution domain).2. SelectionDuring each successive step (called “epoch”), a certain proportion of theexisting population is selected to breed a new generation. Individual solutionsare selected using fitness function, which may be a stochastic functiondefined in order to select a small portion of less fit solutions. This helps21

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!