Hands-On Artificial Intelligence for IoT

Advantages

Genetic algorithms offer some intriguing advantages and can produce results when the tradition gradient-based approaches fail:

• They can be used to optimize either continuous or discrete variables.
• Unlike gradient descent, we do not require derivative information, which also means that there is no need for the fitness function to be continuous and differentiable.
• It can simultaneously search from a wide sampling of the cost surface.
• We can deal with a large number of variables without a significant increase in computation time.
• The generation of the population and calculating their fitness values can be performed in parallel, and hence genetic algorithms are well suited for parallel computers.
• They can work even when the topological ...