2021 IEEE International Intelligent Transportation Systems Conference (ITSC), 2021
The effectiveness of resource allocation under emergencies especially hurricane disasters is cruc... more The effectiveness of resource allocation under emergencies especially hurricane disasters is crucial. However, most researchers focus on emergency resource allocation in a ground transportation system. In this paper, we propose Learningto-Dispatch (L2D), a reinforcement learning (RL) based air route dispatching system, that aims to add additional flights for hurricane evacuation while minimizing the airspace's complexity and air traffic controller's workload. Given a bipartite graph with weights that are learned from the historical flight data using RL in consideration of short-and long-term gains, we formulate the flight dispatch as an online maximum weight matching problem. Different from the conventional order dispatch problem, there is no actual or estimated index that can evaluate how the additional evacuation flights influence the air traffic complexity. Then we propose a multivariate reward function in the learning phase and compare it with other univariate reward designs to show its superior performance. The experiments using the realworld dataset for Hurricane Irma demonstrate the efficacy and efficiency of our proposed schema.
Inspired by De Broglie's pilot wave interpretation of Quantum Mechanics and the subsequent de... more Inspired by De Broglie's pilot wave interpretation of Quantum Mechanics and the subsequent development of Bohm's Quantum Hydrodynamics, we propose a model for the dynamics of a neural network with reaction-diffusion processes described by a modified set of Cohen-Grossberg equations, which we call Neurohydrodynamics. In this approach, a pilot wave interpretation deterministically guides the dynamics of a neural network through the neu-ropotential that arises biologically from reaction-diffusion processes at the synapses of real neurons. We demonstrate that the neuropotential provides a new type of reinforce-ment learning useful for characterizing short-term memory and pattern formation in neural networks, and we compare our results to more traditional reinforcement learning methods through an example. Finally, we discuss extending our approach to include learning, memory, cognition and decision-making processes of the mammalian brain that are often modeled by neural networks.
Uploads
Papers by dahai liu