Dynamic Neighborhood Construction for Structured Large Discrete Action Spaces

arXiv:2305.19891v4 Announce Type: replace
Abstract: Large discrete action spaces (LDAS) remain a central challenge in reinforcement learning. Existing solution approaches can handle unstructured LDAS with up to a few million actions. However, many real-world applications in logistics, production, and transportation systems have combinatorial action spaces, whose size grows well beyond millions of actions, even on small instances. Fortunately, such action spaces exhibit structure, e.g., equally spaced discrete resource units. With this work, we focus on handling structured LDAS (SLDAS) with …

abstract applications arxiv beyond challenge construction cs.ai cs.lg dynamic logistics production reinforcement reinforcement learning solution spaces systems transportation type unstructured world