Reward shaping to improve the performance of deep reinforcement learning in perishable inventory management

Bram J. De Moor*, Joren Gijsbrechts, Robert N. Boute

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

1 Scopus citations

Abstract

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.

Original languageEnglish (US)
JournalEuropean Journal of Operational Research
DOIs
StateAccepted/In press - 2021
Externally publishedYes

Keywords

  • Deep reinforcement learning
  • Inventory
  • Perishable inventory management
  • Reward shaping
  • Transfer learning

ASJC Scopus subject areas

  • Computer Science(all)
  • Modeling and Simulation
  • Management Science and Operations Research
  • Information Systems and Management

Fingerprint

Dive into the research topics of 'Reward shaping to improve the performance of deep reinforcement learning in perishable inventory management'. Together they form a unique fingerprint.

Cite this