Reward shaping to improve the performance of deep reinforcement learning in perishable inventory management
Name:
Publisher version
View Source
Access full-text PDFOpen Access
View Source
Check access options
Check access options
Publication type
Vlerick strategic journal articlePublication Year
2022Journal
European Journal of Operational ResearchPublication Volume
301Publication Issue
2Publication Begin page
535Publication End page
545
Metadata
Show full item recordAbstract
Deep reinforcement learning (DRL) has proven to be an effective, general-purpose technology to develop ‘good’ replenishment policies in inventory management. We show how transfer learning from existing, well-performing heuristics may stabilize the training process and improve the performance of DRL in inventory control. While the idea is general, we specifically implement potential-based reward shaping to a deep Q-network algorithm to manage inventory of perishable goods that, cursed by dimensionality, has proven to be notoriously complex. The application of our approach may not only improve inventory cost performance and reduce computational effort, the increased training stability may also help to gain trust in the policies obtained by black box DRL algorithms.Keyword
Inventory, Perishable Inventory Management, Deep Reinforcement Learning, Reward Shaping, Transfer LearningKnowledge Domain/Industry
Operations & Supply Chain Managementae974a485f413a2113503eed53cd6c53
10.1016/j.ejor.2021.10.045