Complex nonlinear neural network prediction with IOWA layer

Abstract Neural network methods are widely used in business problems for prediction, clustering, and risk management to improving customer satisfaction outcome. The ability of a neural learn complex nonlinear relationship is due its architecture that uses weight parameters transform input data within the hidden layers. Such perform well many situations where ordering inputs simple. However, reordering decision-maker, process not enough get an optimal prediction result. Moreover, existing machine learning algorithms cannot reduce computational complexity by reducing size without losing any information. This paper proposes induced ordered weighted averaging (IOWA) operator artificial IOWA-ANN. reorders according order-inducing variable. proposed sorting mechanism can handle dataset, which results reduced complexities. approach deals with neuron, collects allows degree customisation structure. application further extended IGOWA Quasi-IOWA operators. We present numerical example financial decision-making demonstrate approach's effectiveness handling situations. opens new research area various predictions dataset big enough, such as cloud QoS IoT sensors data. be different learning, networks or hybrid fuzzy other extensions OWA operator.