Lots of objects (computer, smart TV, smartwatch, smartphone, connected objet, ...) belonging to a client can be using the same project. This project, if granted the permission by the client, can evaluate the fitness of chromosomes thus participating in the evolution of the population. It is more convenient to have a centralized online genetic algorithms handling to handle all the chromosome data. This is the purpose of this oga.jod.li initiative.
One important feature of oga.jod.li is the auto optimisation of the genetic algorithm parameters such as the mutation rate, crossover rate, elitism rate and niche parameters through another genetic algorithm. This optimization generally leads to a faster evolution as shown in the parameters documentation https://jod.li/jod-li-initiatives/distribute-optimised-genetic-algorithm-evolution-with-niche-using-oga-jod-li/ .
Niche parameters can improve the general performance of the evolution by avoiding local optima and ensuring a better diversity of the population.
Appart from the mutation, crossover, elitism, niche and automation parameters, the population parameters include chromosome length boundaries (minimum and maximum), chromosome encoding (digit [0-9], alnum [a-zA-Z0-9], alpha [a-zA-Z], lower [a-z], upper [A-Z], hexadecimal [A-F0-9], envelope (json, xml, csv) and population batch size.
Plots allow a comfortable monitoring of the performance of the evolution. Migration allows to copy the best individuals from one population to another.
Start by creating a user. You will be sent a key at your email address. Create then a population with a few parameters. Use the key along with some HTTPS calls to start online genetic algorithm evolution. The server will give new chromosomes to evaluate to your clients and the clients will send the fitness of the chromosomes to the server. Visit this page with your user name and key to view your results. Note that this site does not use any cookie.
Please note that the server can store a maximum of 50 populations per user each storing up to 1000 chromosomes amongst which 100 can be evaluated. Once limits are reached, the least fit chromosomes are deleted to make room for the new ones. Feel free to contact us if you'd like to increase limits.
The maximum length of each chromosome is 3000.
GitHub