Impact of Input Layer of NAR Neural Network on Total Prediction Accuracy  
Author Vladimir Stenchlak


Co-Author(s) Ivan Kuric; Vladimir Tlach; Dariusz Wiecek


Abstract This paper deals with nonlinear autoregressive neural networks. In this paper there is focus on the comparison of two neural network structures for time series prediction. The paper describes how the size of the input layer affects the overall accuracy of the prediction. A data set that is freely available online is used for time series prediction. The data set contains the measured air quality data at moderately frequented road traffic - specifically we focus on carbon monoxide concentrations given in the dataset. The NAR neural network will be in this case a model which can calculate predicted values of carbon monoxide at any time in advance.


Keywords Artificial Intelligence; Neural Networks; Prediction Modelling; Nonlinear Autoregressive Networks, Prediction
    Article #:  RQD26-271

Proceedings of 26th ISSAT International Conference on Reliability & Quality in Design
Virtual Event

August 5-7, 2021