Genetic Programming vs Genetic Algorithms
The genetic algorithm has several disadvantages, for example the length of the strings
are static and limited, it is often hard to describe what the characters of the string means
and the meaning of the characters is limited to whatever they are specified to be. These
problems could perhaps be overcome but a better solution brings back the original
question: How can computers learn to solve problems without being explicitly
programmed? Genetic programming suggests the following solution: simply apply the
genetic operations on computer programs to evolve better and better programs.
There are a lot of frameworks that you can start with Genetic Programming immediately however you do not need them to start using Artificiell Intelligence. You simply need to know about a few core knowledge to solve problems like Traveling Salesman Problem known as TSP in computer world.
There are different methods to select chromosomes from a population to be parents for crossover to next generation.
Roulette wheel selection
Boltzman selection
Tournament selection
Rank selection
Steady state selection
Crossover and Mutation Probability
This is two parameters of Genetic Algorithms.
Crossover takes place once the chromosomes are selected based on one of the methods above.
This will produce offsprings. A crossover of 100% or 1.0 means that all of the chromosomes will be used for reproduction, i.e. there is no servivors. Empirical studies shows that a Crossover probability of between 0.65 and 0.85 will achieve better results.If it is 0% nothing is changed and the offspring is a exact copy of one of the parent.
Some researchers recommend decreasing the cross-over probability and increasing the mutation probability in depending on the number of performed generations. Others say that it is good if the genetic algorithm has a dynamic cross-over probability that changes in dependence to spread value inside the selection mechanism.
Resources:
http://en.wikipedia.org/wiki/Genetic_algorithm
Crossover and Mutation Probability
This is two parameters of Genetic Algorithms.
Crossover takes place once the chromosomes are selected based on one of the methods above.
This will produce offsprings. A crossover of 100% or 1.0 means that all of the chromosomes will be used for reproduction, i.e. there is no servivors. Empirical studies shows that a Crossover probability of between 0.65 and 0.85 will achieve better results.If it is 0% nothing is changed and the offspring is a exact copy of one of the parent.
Some researchers recommend decreasing the cross-over probability and increasing the mutation probability in depending on the number of performed generations. Others say that it is good if the genetic algorithm has a dynamic cross-over probability that changes in dependence to spread value inside the selection mechanism.
Resources:
http://en.wikipedia.org/wiki/Genetic_algorithm
