Abstract: This paper presents an improved deep reinforcement learning-based (DRL) approach for end-to-end models using a Graph Neural Network(GNN). The proposed method aims to improve end-to-end deep ...