Genetic algorithms and machine learning
The practical purpose of a genetic algorithm as an optimization technique is to solve problems by finding the most relevant or fittest solution among a set or group of solutions. Genetic algorithms have many applications in machine learning, which are as follows:
Discrete model parameters: Genetic algorithms are particularly effective in finding the set of discrete parameters that maximizes the log likelihood. For example, the colorization of a black and white movie relies on a large but finite set of transformations from shades of grey to the RGB color scheme. The search space is composed of the different transformations and the objective function is the quality of the colorized version of the movie.
Reinforcement learning: Systems that select the most appropriate rules or policies to match a given dataset rely on genetic algorithms to evolve the set of rules over time. The search space or population is the set of candidate rules, and the objective...
