Inventory management of new products in retailers using model-based deep reinforcement learning

Tsukasa Demizu, Yusuke Fukazawa, Hiroshi Morita

Abstract: Highlights•An inventory management problem with demand uncertainty for new products is studied.•This problem can be addressed with model-based deep reinforcement learning.•The proposed method combines offline model learning and online supply planning.•Simulations using real-world datasets show superior inventory control performance.