Recurrent Neural Network based Aquila optimizer on traffic images

Abstract

Forecasting traffic flow and maintaining a smooth flow of traffic in smart cities is necessary. Therefore, a better image prediction technique must be implemented in traffic with better accuracy and precision for an improved decision-making system. Better image prediction techniques can be attained by implementing optimizations and neural networks. By this review, optimizations such as Whale optimization algorithm, Beluga whale optimization, Grey wolf optimizer, Focused ant colony optimization, Improved artificial bee colony algorithm, Spotted hyena optimizer simulated annealing, Honey badger algorithm, Monarch butterfly optimization, Siberian tiger optimization, and Chaotic Ant lion optimization are studied with their search mechanics. Modified Aquila Optimization (AO) and Recurrent Neural Networks (RNN) are studied in detail with their accuracy, precision, recall, and sensitivity. Furthermore, this investigation explains the merits and demerits of hybrid AO models and RNN models. Additionally, this review shows the correlation between ANN, CNN, and RNN. The functioning assessment of the optimization and neural networks are also investigated. The AO has an accuracy of 99.76%, and the RNN has an accuracy of 99.98%. The recall percentage of RNN is 98.97% and the sensitivity of the AO is 99.69%. Generally, this paper provides a brief overview of the feasible approaches for implementing enhanced optimization techniques in various fields. In consideration of this, future research works can be able to implement modified optimization models in traffic image prediction systems.