Attentive TabNet Enabled Collision Prediction for V2X Network

Le résumé fourni par la source

Collision prediction is crucial for safety in Vehicle-to-Everything (V2X) networks, essential for smart cities and intelligent transportation. Traditional methods struggle with complexity, while learning-based approaches face challenges in feature extraction and prediction accuracy. To address this, we propose attentive TabNet for collision prediction in V2X networks. Our model utilizes attention mechanisms in deep neural networks to focus on key spatial and temporal features, enhancing prediction reliability. We also apply data preprocessing to handle imbalances. Experimental evaluations using real-world V2X data show that our method outperforms random forest and neural network-based algorithms in accuracy and efficiency. Both simulations and real-world tests confirm its effectiveness. By improving collision prediction, our approach enhances road safety, advancing intelligent transportation and smart city initiatives.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.