Open 11Johan11 opened 2 hours ago
Something we could experiment in later stages if we have time is to have "Self-Adaptive" behaviour where we include the mutation parameters in the solution itself, so in the beginning it will automatically explore heavy(mutate aggressively) but in later stages exploit heavy(mutate very little)
Takes a solution and mutates it, For our permutation based solutions we can choose from
For Genetic Algorithms it's always about striking a balance between exploration and exploitation, Exploration (looking through the search space and finding a optimum) Exploitation (looking through closely at a optimum trying to reach the peak)
Exploration is usually very important in the beginning where diversity is key
We could implement all of them and experiment & evaluate.