C# 클래스 SharpNeat.Core.EvaluationInfo

Encapsulates genome evaluation data; specifically whether the genome has been evaluated, the genome's fitness, how many times the genome has been evaluated and if more than once then the mean fitness. Mean fitness is sometimes used when evaluations are non-determininstic (where each evaluation with the same parameters results in a different fitness, usually following some distribution curve) or else successive evaluations use different evaluation parameters and therefore result in different fitnesses. This approach is sometimes used in order to evaluate against a very computationally expensive fitness scheme, or indeed a computationally intractable scheme - i.e. we only execute against some subset of the problem space for each genome evaluation, and thus we obtain a more representative value for a genome's fitness by averaging over successive evaluations.