深度强化学习（DRL）在库存管理中的应用是一个新兴领域。然而，传统DRL算法最初是为游戏博弈和机器人等多样化领域开发的，可能并不适合库存管理所带来的特定挑战。因此，这些算法往往无法超越既有的启发式方法；例如，在损失销售库存控制中，尚无现有的DRL方法能够持续优于有上限的基础库存策略。这凸显了DRL在库存管理实际应用中的一个关键空白：库存问题的高度随机性需要量身定制的解决方案。为此，我们提出了深度受控学习（DCL），这是一种专为高度随机问题设计的新型DRL算法。DCL基于近似策略迭代，并结合了一种高效的仿真机制，将逐次减半（Sequential Halving）与公共随机数（Common Random Numbers）相结合。我们的数值研究表明，DCL在多种库存场景下均能持续优于最先进的启发式方法和DRL算法，包括损失销售、易腐库存系统以及具有随机提前期的库存系统。DCL在所有测试用例中均实现了更低的平均成本，同时保持不超过0.2%的最优性差距。值得注意的是，这一性能是在所有实验中使用同一组超参数实现的，这凸显了我们方法的鲁棒性和泛化能力。这些发现有助于持续推进面向库存管理的定制化DRL算法的探索，为该领域的进一步研究和实际应用奠定了基础。
英文摘要：
The application of Deep Reinforcement Learning (DRL) to inventory management is an emerging field. However, traditional DRL algorithms, originally developed for diverse domains such as game-playing and robotics, may not be well-suited for the specific challenges posed by inventory management. Consequently, these algorithms often fail to outperform established heuristics; for instance, no existing DRL approach consistently surpasses the capped base-stock policy in lost sales inventory control. This highlights a critical gap in the practical application of DRL to inventory management: the highly stochastic nature of inventory problems requires tailored solutions. In response, we propose Deep Controlled Learning (DCL), a new DRL algorithm designed for highly stochastic problems. DCL is based on approximate policy iteration and incorporates an efficient simulation mechanism, combining Sequential Halving with Common Random Numbers. Our numerical studies demonstrate that DCL consistently outperforms state-of-the-art heuristics and DRL algorithms across various inventory settings, including lost sales, perishable inventory systems, and inventory systems with random lead times. DCL achieves lower average costs in all test cases while maintaining an optimality gap of no more than 0.2\%. Remarkably, this performance is achieved using the same hyperparameter set across all experiments, underscoring the robustness and generalizability of our approach. These findings contribute to the ongoing exploration of tailored DRL algorithms for inventory management, providing a foundation for further research and practical application in this area.
DeepStock: Reinforcement Learning with Policy Regularizations for Inventory Management
Booth School of Business, University of Chicago（芝加哥大学商学院）；Taobao & Tmall Group（淘宝与天猫集团）；School of Economics, Sichuan University（四川大学经济学院）；Graduate School of Business, Columbia University（哥伦比亚大学商学院）；School of Operations Research and Information Engineering, Cornell University（康奈尔大学运筹学与信息工程学院）
InvEvolve: Evolving White-Box Inventory Policies via Large Language Models with Performance Guarantees
Shanghai University of Finance and Economics（上海财经大学）；Shanghai Jiao Tong University（上海交通大学）；The Chinese University of Hong Kong, Shenzhen（香港中文大学（深圳））；Tsinghua University（清华大学）；Boston College（波士顿大学）
ORPR: An OR-Guided Pretrain-then-Reinforce Learning Model for Inventory Management
Department of Industrial Engineering, Tsinghua University（清华大学工业工程系）；Supply Chain Tech Team Y, JD.com（京东供应链技术团队）；Faculty of Engineering and Faculty of Business and Economics, The University of Hong Kong（香港大学工程学院和商学院）
Hard Constraints, Smooth Gradients: Learning Feasible Inventory Policies via Differentiable Projection
TUM School of Management, Technical University of Munich（慕尼黑工业大学TUM管理学院）；Esade, Ramon Llull University（拉蒙·柳利大学ESADE商学院）；Munich Data Science Institute (MDSI), Technical University of Munich（慕尼黑工业大学慕尼黑数据科学研究所）
Control of Dual-Sourcing Inventory Systems using Recurrent Neural Networks
Frankfurt School of Finance and Management（法兰克福金融管理学院）；University of California, Los Angeles（加利福尼亚大学洛杉矶分校）；University of Zurich（苏黎世大学）；Erasmus University Rotterdam（鹿特丹伊拉斯姆斯大学）
AI Agents for Inventory Control: Human-LLM-OR Complementarity
Stern School of Business, New York University（纽约大学斯特恩商学院）；Columbia University（哥伦比亚大学）；Graduate School of Business and Data Science Institute, Columbia University（哥伦比亚大学商学院与数据科学研究院）
Predictive and Prescriptive Performance of Bike-Sharing Demand Forecasts for Inventory Management
Technical University of Denmark（丹麦技术大学）；Technical University of Munich（慕尼黑工业大学）；University of Groningen（格罗宁根大学）；University of Twente（特文特大学）；Munich Data Science Institute（慕尼黑数据科学研究所）
Recurrent U-net: Deep learning to predict daily summertime ozone in the United States
University of Toronto（多伦多大学）；Carnegie Mellon University（卡内基梅隆大学）；California Institute of Technology（加州理工学院）；University of Science and Technology of China（中国科学技术大学）；National Center for Atmospheric Research（美国国家大气研究中心）
Disentangling Attention in Deep Operator Learning: A Controlled Study of Data-Driven and Physics-Informed Architectures
National Center for Supercomputing Applications, University of Illinois at Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校国家超级计算应用中心）；Department of Mechanical Engineering, Stanford University（斯坦福大学机械工程系）
Deep Meditations: Controlled navigation of latent space
Department of Computing, Goldsmiths University of London（伦敦大学金史密斯学院计算机系）；Creative Computing Institute, University of the Arts, London（伦敦艺术大学创意计算学院）
我们介绍了一种方法，允许用户创造性地探索和导航深度生成模型的巨大潜在空间。具体来说，我们的方法使用户\textit{discover}和\textit{design} \textit{trajectories}在这些高维空间，构建故事，并产生基于时间的媒体，如视频- \textit{with meaningful control over narrative}。我们的目标是鼓励和帮助使用深度生成模型作为创造性表达和讲故事的媒介，并进行有意义的人类控制。我们的方法类似于传统的视频制作管道，因为我们使用传统的非线性视频编辑器与代理剪辑，并符合潜在空间向量的数组。在\url{http://deepmeditations.ai}中可以看到示例。
Acquisition-Independent Deep Learning for Quantitative MRI Parameter Estimation using Neural Controlled Differential Equations
Amsterdam University Medical Center（阿姆斯特丹大学医学中心）；Cancer Center Amsterdam（阿姆斯特丹癌症中心）；University of Queensland（昆士兰大学）；UNSW Sydney（悉尼新南威尔士大学）；University of Amsterdam（阿姆斯特丹大学）；University of Fribourg（弗里堡大学）；University of Bern（伯尔尼大学）
