DeepStock: Reinforcement Learning with Policy Regularizations for Inventory Management
This work addresses hyperparameter tuning challenges in DRL for inventory management, offering incremental improvements for e-commerce platforms like Alibaba.
The paper tackles the problem of high sensitivity to hyperparameters in Deep Reinforcement Learning (DRL) for inventory management by imposing policy regularizations based on classical inventory concepts like 'Base Stock', resulting in accelerated hyperparameter tuning and improved performance, as demonstrated in a 100% deployment on Alibaba's Tmall platform.
Deep Reinforcement Learning (DRL) provides a general-purpose methodology for training inventory policies that can leverage big data and compute. However, off-the-shelf implementations of DRL have seen mixed success, often plagued by high sensitivity to the hyperparameters used during training. In this paper, we show that by imposing policy regularizations, grounded in classical inventory concepts such as "Base Stock", we can significantly accelerate hyperparameter tuning and improve the final performance of several DRL methods. We report details from a 100% deployment of DRL with policy regularizations on Alibaba's e-commerce platform, Tmall. We also include extensive synthetic experiments, which show that policy regularizations reshape the narrative on what is the best DRL method for inventory management.