Learning General Inventory Management Policy for Large Supply Chain Network

TL;DR

This paper introduces a reinforcement learning algorithm for large-scale supply chain inventory management, effectively handling the complexity of numerous products and retailers through approximate simulation.

Contribution

It presents a novel RL-based approach with approximate simulation to manage large supply chain networks, addressing computational challenges of previous methods.

Findings

Successfully manages large supply chain networks with many products and retailers.

Demonstrates effectiveness on both real and artificial data.

Outperforms traditional algorithms in handling complexity.

Abstract

Inventory management in warehouses directly affects profits made by manufacturers. Particularly, large manufacturers produce a very large variety of products that are handled by a significantly large number of retailers. In such a case, the computational complexity of classical inventory management algorithms is inordinately large. In recent years, learning-based approaches have become popular for addressing such problems. However, previous studies have not been managed systems where both the number of products and retailers are large. This study proposes a reinforcement learning-based warehouse inventory management algorithm that can be used for supply chain systems where both the number of products and retailers are large. To solve the computational problem of handling large systems, we provide a means of approximate simulation of the system in the training phase. Our experiments on both real and artificial data demonstrate that our algorithm with approximated simulation can successfully handle large supply chain networks.