Smart Transportation Without Neurons -- Fair Metro Network Expansion with Tabular Reinforcement Learning
Researchers have developed a tabular reinforcement learning approach for optimizing metro network expansion, outperforming deep reinforcement learning in efficiency and interpretability. This method incorporates social equity criteria into its reward functions, balancing efficiency with fairness. Tested in real-world scenarios in Xi'an and Amsterdam, the approach significantly reduced training episodes and carbon emissions while maintaining competitive performance. AI
IMPACT Offers a more efficient and interpretable method for complex optimization problems, potentially applicable beyond transportation.