Multilayer feedforward networks are universal approximators
Neural Networks
Learning internal representations by error propagation
Parallel distributed processing: explorations in the microstructure of cognition, vol. 1
Artificial Neural Networks: Learning Algorithms, Performance Evaluation, and Applications
Artificial Neural Networks: Learning Algorithms, Performance Evaluation, and Applications
| Hi-index | 0.00 |
The paper demonstrates an artificial intelligence method known as the Artificial Neural Network (ANN) approach based on supervised neural networks to estimate the number of persons fatally injured in motor vehicle accidents. In order to analyze a data set related to fatal accidents, the Artificial Neural Network Estimating Fatal Accident (ANNEFA) model is developed by using social and traffic-related variables, population and motor-vehicle registrations. To obtain the best form of ANNEFA, various ANN models having different transfer functions, different number of neurons and different train algorithms are designed. The ANNEFA model formed with fourteen neurons, tansig transfer function and Levenberg-Marquardt training algorithm provides the best fit to training and test data. Fluctuations in variables used in historical data are reflected in results of the ANNEFA model. Estimates of the ANNEFA model compared with the results of Revised Smeed Equation (RSE) reconstituted from Smeed Equation in accord with the USA Data Set. Thus, this study provides a benchmark for predicting fatality in motor-vehicle accidents in the form of a numerical and graphical comparison between results of the ANNEFA and RSE. The results indicate that the ANN model is a proper approach in predicting fatalities in motor-vehicle crashes.