A large challenge in Artificial Intelligence (AI) is training control agents that can properly adapt to variable environments. Environments in which the conditions change can cause issues for agents trying to operate in them. Building algorithms that can train agents to operate in these environments and properly deal with the changing conditions is therefore important. NeuroEvolution of Augmenting Topologies (NEAT) was a novel Genetic Algorithm (GA) when it was created, but has fallen aside with newer GAs outperforming it. This paper furthers the research on this subject by implementing various versions of improved NEAT in a variable environment to determine if NEAT can perform well in these environments. The improvements included, in every combination, are: recurrent connections, automatic feature selection, and increasing population size. The recurrent connections improvement performed extremely well. The automatic feature selection improvement was found to be detrimental to performance, and the increasing population size improvement lowered performance a small amount, but decreased computation requirements noticeably.

There have been recent efforts to move to population-based structural health monitoring (PBSHM) systems. One area of PBSHM which has been recognised for potential development is the use of multi-task learning (MTL); algorithms which differ from traditional independent learning algorithms. Presented here is the use of the MTL, ''Joint Feature Selection with LASSO'', to provide automatic feature selection for a structural dataset. The classification task is to differentiate between the port and starboard side of a tailplane, for samples from two aircraft of the same model. The independent learner produced perfect F1 scores but had poor engineering insight; whereas the MTL results were interpretable, highlighting structural differences as opposed to differences in experimental set-up.

There's a couple of reasons why you might want to feature selection. One is to avoid overfitting and get a better model. In practice, I have rarely seen that happen. It's not usually what I would try to do to increase performance. If I'm only interested in performance I probably would not try to do automatic feature selection unless I think only a very small subset of my feature is actually important.