The basics of deep learning algorithms and their effect on driving behavior and vehicle communications

Deep learning (DL) is now considered one of the most important fields in computer science for its capability of delivering the best results from enormous quantities of data and really complex information. When it comes to intelligent vehicle networks, deep learning can play an important factor in improving the communication between vehicles and improving road safety conditions. The analysis of vehicle communication is the analysis of all available data and features from the vehicles in order to decide on the best outcome and traffic route for a road-based scenario. This operation can seem complicated and challenging, especially when dealing with the unpredictability and the inescapable human error factor in the decision-making process that still interferes with the roads and traffic structures of our days. However, the evolution of deep learning algorithms is undeniably an exciting prospect to deal with these challenges, from the basic principles of artificial neural networks, convolutional neural networks, and recurrent neural networks to the newest state-of-the-art solutions of variational auto-encoders, generative adversarial networks, and transformers. These principles alongside the variant nuance architectures combined with the powerful computing powers available in our time are on the edge of delivering the futuristic vision of totally autonomous vehicles and road networks that we thought is still too far to achieve. In this chapter, we are going to discuss the basics of deep learning algorithms, the pros and cons of every method and the pre-processing steps and hyperparameters refinement. Then we are going to describe and elaborate more with some examples the effect of these principles on driving behavior and vehicle communication.