Artificial Intelligence for Resource Allocation Tasks
MCML Authors
Niklas Strauß
Dr.
Abstract
Niklas Strauß
Dr.
Abstract
This thesis presents deep reinforcement learning approaches for complex resource allocation tasks, including discrete, continuous, and resource collection problems. It introduces novel neural architectures achieving state-of-the-art results in spatial resource allocation, multi-agent collection, and dynamic ambulance redeployment, including electric ambulances. For continuous tasks like portfolio optimization, it proposes efficient methods to handle allocation constraints, ensuring compliance during training and deployment. (Shortened).
BibTeXKey: Str24