Artificial intelligence

Abstract: The paper addresses the problem of adaptive inventory management under conditions of unstable demand, high sales variability, and the need for rapid decision-making in logistics systems. An analysis of contemporary approaches to the application of machine learning and reinforcement learning methods for inventory management support is conducted. It is established that most existing solutions are focused either on improving demand forecasting accuracy followed by optimization of inventory replenishment parameters, or on the use of end-to-end intelligent models in which system state analysis and decision-making are implemented within a single software and algorithmic component. A conceptual architecture for adaptive inventory replenishment policy selection is proposed. The architecture is based on the principle of functional separation between the processes of evaluating alternative replenishment policies and making replenishment decisions. A distinctive feature of the proposed approach is the use of an evaluator-selector architecture, in which one intelligent component estimates the expected effectiveness of alternative replenishment policies, while another component performs their adaptive selection according to the current state of the system. It is demonstrated that improving demand forecasting accuracy does not necessarily guarantee the selection of the most effective inventory replenishment policy, since management performance is determined by a combination of economic, logistical, and operational factors. Unlike most contemporary machine learning- and reinforcement learning-based inventory management systems, the proposed approach provides architectural modularity, improves the interpretability of the decision-making process, facilitates integration with enterprise information systems, and enables the independent enhancement of individual functional components. The scientific novelty of the study lies in the development of conceptual foundations for an adaptive inventory replenishment policy selection architecture based on the functional separation of policy evaluation and decision-making processes. The practical significance of the research is associated with the possibility of using the proposed approach as a foundation for intelligent decision support systems for inventory management under conditions of uncertain demand and a dynamic market environment.

Keywords: adaptive inventory management; adaptive policy selection; machine learning; reinforcement learning; intelligent decision support system; evaluator-selector architecture.

References:

1. Silver E. A., Pyke D. F., Thomas D. J. Inventory and Production Management in Supply Chains. 4th ed. Boca Raton : CRC Press, 2016. 664 p. DOI: 10.1201/b10463.
2. Mohammed Z., Chafi A., El Hammoumi M. A hybrid learning framework for forecasting uncertainty and adaptive inventory planning in retail supply chains // Supply Chain Analytics. 2025. Vol. 13. Article 100180. DOI: 10.1016/j.sca.2025.100180.
3. Rolf B., Jackson I., Müller M., Lang S., Reggelin T., Ivanov D. A review on reinforcement learning algorithms and applications in supply chain management // International Journal of Production Research. 2023. Vol. 61, No. 20. P. 7151–7179. DOI: 10.1080/00207543.2022.2140221.
4. Boute R. N., Gijsbrechts J., van Jaarsveld W., Vanvuchelen N. Deep reinforcement learning for inventory control: A roadmap // European Journal of Operational Research. 2022. Vol. 298, No. 2. P. 401–412. DOI: 10.1016/j.ejor.2021.07.016.
5. Cuartas C., Aguilar J. Hybrid algorithm based on reinforcement learning for smart inventory management // Journal of Intelligent Manufacturing. 2023. Vol. 34. P. 123–149. DOI: 10.1007/s10845-022-01982-5.
6. Bergsma R., de Ruijt C., Bhulai S. A systematic review of machine learning approaches in inventory control optimization // Operations Research Perspectives. 2025. Vol. 15. Article 100367. DOI: 10.1016/j.orp.2025.100367.
7. Muth M., Lingenfelder M., Nufer G. The application of machine learning for demand prediction under macroeconomic volatility: a systematic literature review // Management Review Quarterly. 2025. Vol. 75. P. 2759–2802. DOI: 10.1007/s11301-024-00447-8.
8. Samal T., Ghosh A. Ensemble-based predictive analytics for demand forecasting in multi-channel retailing // Expert Systems with Applications. 2026. Vol. 299, Part D. Article 130212. DOI: 10.1016/j.eswa.2025.130212.
9. Wang J., Luo L. A forecasting method of multi-category product sales: analysis and application // Management System Engineering. 2023. Vol. 2. Article 2. DOI: 10.1007/s44176-023-00012-9.
10. Liu Y., Kalaitzi D., Wang M., Papanagnou C. A machine learning approach to inventory stockout prediction // Journal of Digital Economy. 2025. Vol. 4. P. 144–155. DOI: 10.1016/j.jdec.2025.06.002.
11. Goulart D. D., de Carvalho R. B., Henriques M. A., Nunes Carvalho B. K. G. Predictive models for inventory optimization: a machine learning application for demand forecasting at a construction supplies distributor // Future Business Journal. 2026. Vol. 12. Article 92. DOI: 10.1186/s43093-026-00807-8.
12. Bansal V., Bisi A., Roy D. Integrated inventory replenishment and online demand allocation decisions for an omnichannel retailer with ship-from-store strategy // European Journal of Operational Research. 2024. Vol. 316. P. 1085–1100. DOI: 10.1016/j.ejor.2024.02.027.
13. Ren X. et al. Data-driven analysis on inventory problem for anticipatory shipping // Computers & Industrial Engineering. 2025. Article 111038. DOI: 10.1016/j.cie.2025.111038.
14. Vieira B. P., Frazzon E. M. Application of Reinforcement Learning to Improve Finished Goods Inventory Management: A Systematic Review // IFAC-PapersOnLine. 2025. Vol. 59, No. 10. P. 841–846. DOI: 10.1016/j.ifacol.2025.09.143.
15. Geevers K., van Hezewijk L., Mes M. R. K. Multi-echelon inventory optimization using deep reinforcement learning // Central European Journal of Operations Research. 2024. Vol. 32. P. 653–683. DOI: 10.1007/s10100-023-00872-2.
16. Gijsbrechts J., Boute R. N., Van Mieghem J. A., Zhang D. J. Can Deep Reinforcement Learning Improve Inventory Management? Performance on Lost Sales, Dual-Sourcing, and Multi-Echelon Problems // Manufacturing & Service Operations Management. 2022. Vol. 24, No. 3. P. 1349–1368. DOI: 10.1287/msom.2021.1064.
17. Park H., Choi D. G., Min D. Adaptive inventory replenishment using structured reinforcement learning by exploiting a policy structure // International Journal of Production Economics. 2023. Vol. 266. Article 109029. DOI: 10.1016/j.ijpe.2023.109029.
18. Dehaybe H., Catanzaro D., Chevalier P. Deep Reinforcement Learning for inventory optimization with non-stationary uncertain demand // European Journal of Operational Research. 2024. Vol. 314, No. 2. P. 433–445. DOI: 10.1016/j.ejor.2023.10.007.
19. Tian R., Lu M., Wang H., Wang B., Tang Q. IACPPO: A deep reinforcement learning-based model for warehouse inventory replenishment // Computers & Industrial Engineering. 2024. Vol. 187. Article 109829. DOI: 10.1016/j.cie.2023.109829.
20. Zhang Y., He L., Zheng J. A Deep Reinforcement Learning-Based Dynamic Replenishment Approach for Multi-Echelon Inventory Considering Cost Optimization // Electronics. 2025. Vol. 14, No. 1. Article 66. DOI: 10.3390/electronics14010066.
21. Rizqi Z. U., Chou S.-Y. Neuroevolution reinforcement learning for multi-echelon inventory optimization with delivery options and uncertain discount // Engineering Applications of Artificial Intelligence. 2024. Vol. 134. Article 108670. DOI: 10.1016/j.engappai.2024.108670.
22. Demizu T., Fukazawa Y., Morita H. Inventory management of new products in retailers using model-based deep reinforcement learning // Expert Systems with Applications. 2023. Vol. 229. Article 120256. DOI: 10.1016/j.eswa.2023.120256.
23. Teck S., Phạm T.-S., Rousseau L.-M., Vansteenwegen P. Deep reinforcement learning for the real-time inventory rack storage assignment and replenishment problem // European Journal of Operational Research. 2025. Vol. 327, No. 2. P. 606–622. DOI: 10.1016/j.ejor.2025.05.008.
24. Kolyaei M., Zhang L., Blom M. Inventory replenishment and fulfilment decisions for an omnichannel retailer: a reinforcement learning-based method // International Journal of Production Research. 2025. Vol. 63, No. 24. P. 9571–9592. DOI: 10.1080/00207543.2025.2520596.
25. An L., Li A. A., Moseley B., Ravi R. The Nonstationary Newsvendor with (and Without) Predictions // Manufacturing & Service Operations Management. 2025. Vol. 27, No. 3/4. DOI: 10.1287/msom.2024.1168.
26. Zhang W., Li C., Qin H., Xu Y., Zhu R. Thompson Sampling for Repeated Newsvendor // CoRR/arXiv. 2025. DOI: 10.48550/arXiv.2502.09900.