Hybrid Neuro Genetic Causal Convolution Based Autoencoders for Traffic Prediction in Smart Cities

This research presents a novel approach to traffic prediction in smart cities using Hybrid Neuro-Genetic Causal Convolution based Autoencoders (HNG-CCA). Urban traffic congestion has become a significant challenge in modern urban planning, necessitating accurate and efficient predictive models. In this work, we propose a hybrid architecture that combines neuro-genetic techniques with causal convolutional autoencoders to enhance the predictive capabilities of traffic patterns. The neuro-genetic approach optimizes the autoencoder architecture, while the causal convolutional layers capture the temporal dependencies inherent in traffic data. Our experiments on real-world traffic datasets demonstrate that the HNG-CCA outperforms existing methods in terms of prediction accuracy and generalization. This hybrid approach not only contributes to the field of traffic prediction but also showcases the potential of combining diverse machine learning paradigms to address complex urban challenges in smart cities.