Clustering techniques using Genetic Algorithms and Particle Swarm Optimization
A genetic algorithm is based on Darwin's ideas of evolution. Basically, it takes a population of n individuals, initializes them as possible solutions to a problem, and through crossovers, mutations, and sometimes reproductions, evolves the population until some condition is satisfied.
In this approach, the algorithm aims to discover coordinates to some clusters. It assumes prior knowledge of the problem, i.e., the number of clusters. The population will be of n individuals, aimed to find the coordinates, with some probability of crossover and mutation (we recommend 0.85 and 0.01) until some given number of iterations. The data will be provided through an input file.
$ python Genetic.py n_individuals n_clusters p_crossover p_mutation iterations input_file
PSO is a search algorithm that is based on birds behavior. A particle will be part of a population, and using its best position so far (cognitive factor), the best position in its neighbourhood (social factor), and its 'disposition to move' (inertia), it will move on the solution space looking for the place of minimum error.
The implementation of the algorithm uses a maximum and a minimum value for inertia (we suggest from 0.9 and 0.5), and cognitive and social factors are suggested to be 1.50 each. The input file will hold the data.
$ python PSO.py n_particles n_clusters inertia_max inertia_min fcognitive fsocial iterations input_file