In this paper, we have presented a new approach to selecting features for neural nets called
ANNIGMA-wrapper that makes the wrapper model of feature selection practical in real-world
neural net-based applications. The ANNIGMA-wrapper approach was successfully applied to two
real-world data mining applications in the helicopter industry. Experimental results against standard datasets from UCI repository show that ANNIGMA-wrapper performs well for datasets with
various characteristics and suggest that a simple approach like ANNIGMA-wrapper can perform
equally well or better than existing sophisticated feature selection techniques.