Distributional reinforcement learning for inventory management in multi-echelon supply chains

Wu, Guanglei; Servia, Miguel Ángel de Carvalho; Mowbray, Max

doi:10.1016/j.dche.2022.100073

Cited by 7 publications

(6 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In this area, machine learning has been used from different perspectives. In the studies by Wang et al (2023) and Wu et al (2023) two distinct but complementary approaches are presented for the use of reinforcement learning (RL) in inventory management. While Wang et al (2023) focuses on lot sizing for perishable materials in an uncertain environment, Wu et al (2023) introduces a risk management approach in its objective function to balance risky decisions.…”

Section: Inventory Managementmentioning

confidence: 99%

“…In the studies by Wang et al (2023) and Wu et al (2023) two distinct but complementary approaches are presented for the use of reinforcement learning (RL) in inventory management. While Wang et al (2023) focuses on lot sizing for perishable materials in an uncertain environment, Wu et al (2023) introduces a risk management approach in its objective function to balance risky decisions. Both studies highlight the importance of prediction in defining the state of the reinforcement learning model and the need for real-time sequential decision-making for inventory management.…”

Section: Inventory Managementmentioning

confidence: 99%

“…• Structural Equation Model (SEM) validates the factors that define the variables used by the Bayesian Network (BN) model (Kamble et al, 2021) • Voting Analytical Hierarchy Process (VAHP) calculates the relative weights of criteria that are then used in BN Bayesian Networks (Dohale et al, 2022) Perspective 2. When results from other models or techniques feed machine learning • Adaptive Neuro-Fuzzy Inference System (ANFIS) integrates with time series analysis to predict order arrival rate (Leung et al, 2020) • Recurrent neural network (RNN) parameterizes the policy of a Distributive Reinforcement Learning (RL) Framework and optimization models explore the control space and state (Wu et al, 2023) Demand forecast • The Machine Learning model maintains and synthesizes the criteria scores generated by the Best-Worst Fuzzy Method (BWM) and the Fuzzy Inference System (FIS) (Alavi et al, 2021) Logistics and Transport…”

Section: New Technologies Adoptionmentioning

confidence: 99%

“…Likewise, vertical, and horizontal integration stands out as an imperative need. Supply chain decisions cannot be isolated from the strategic and tactical decisions of the organization (Wu et al, 2023). In addition, we face the challenge of incorporating additional uncertainties, such as supply disruptions or quality problems, into different models and algorithms.…”

Section: Further Research Opportunitiesmentioning

confidence: 99%

See 3 more Smart Citations

Integration of machine learning in the supply chain for decision making: A systematic literature review

Polo-Triana,

Gutierrez,

Leon-Becerra

2024

JIEM

View full text Add to dashboard Cite

Purpose: This study presents a systematic literature review that provides a broad and holistic view of how machine learning can be used and integrated to enhance decision-making in various areas of the supply chain, highlighting its combination with other techniques and models.Design/methodology/approach: An exhaustive literature review used three sets of keywords in the Scopus and Web of Science (WoS) databases. Through a rigorous filtering process, 70 articles were selected from an initial total of 410, focusing on those that specifically addressed the intersection of machine learning and decision-making in supply chain management.Findings: Machine learning has proven to be an essential tool in the supply chain, with applications in inventory management, logistics, and transportation, among others. Its integration with other techniques has led to significant advances in decision-making, improving efficiency in complex environments. Combining machine learning methods with traditional techniques has been particularly effective, and integration with emerging technologies has opened up new application possibilities.Originality/value: Unlike previous studies that focused on specific areas, this study offers a broad perspective on the application of machine learning in the supply chain. Additionally, combining machine learning techniques with other models is highlighted, representing added value for the scientific community and suggesting new avenues for future research.

show abstract

Section: Inventory Managementmentioning

confidence: 99%

Section: Inventory Managementmentioning

confidence: 99%

Section: New Technologies Adoptionmentioning

confidence: 99%

Section: Further Research Opportunitiesmentioning

confidence: 99%

See 2 more Smart Citations

Integration of machine learning in the supply chain for decision making: A systematic literature review

Polo-Triana,

Gutierrez,

Leon-Becerra

2024

JIEM

View full text Add to dashboard Cite

show abstract

“…The structure of data-driven models is exible and can be promptly adapted to different variables or processes. They are faster to evaluate than mechanistic models, 12 making them useful in real-time simulation, [13][14][15][16] optimization, [17][18][19][20] and so sensor development. [21][22][23] However, since no physical knowledge is used, their extrapolatory abilities are oen limited, and their performance depends on the quantity and quality of data available, which might classify their usage in certain scenarios as unsafe.…”

Section: Introductionmentioning

confidence: 99%