Scholarlink

<p>Meeting consumer demand while reducing stock-outs and surplus inventory requires effective inventory management. Nonlinear relations between demand patterns, logistics factors and external influences, however, are complicated and cannot be addressed by traditional inventory systems and shallow learning models. In order to overcome this problem, this research offers a smart inventory management system based on a Dense Nested Attention Network (DNAN) to classify states of inventory. Pre-processing of inventory information based on the sales history, logistics, and contextual information is performed by a Grid-Constrained Data Cleansing algorithm and then the most effective feature selection is done using the Electric Eel Foraging Optimization Algorithm (EeFOA). The DNAN model takes advantage of dense connectivity and hierarchical attention to effectively represent the temporal trend of demand and capture the contextual dependence of demand and stock-out risks to classify the level of demand and stock-out risk correctly. Experimental results demonstrate that the proposed DNAN-based approach achieves 99% classification accuracy, outperforming other attention-based neural models in both accuracy and computational efficiency. This work’s primary contribution is to show how dense connection and nested attention enhance decision-oriented inventory classification, providing a workable and scalable intelligent inventory control solution.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A smart inventory system utilizing gated axial attention for improved accuracy and customer satisfaction

  • 期刊论文

  • 发表日期：

    2026年2月9日

查看全文
  • S. Thanga Helina,
  • Julia Punitha Malar Dhas,
  • C. P. Shirley,
  • E. Rushit Gnanaroy,
  • P. Joyce Beryl Princess,
  • M. B. Abisha

摘要

Meeting consumer demand while reducing stock-outs and surplus inventory requires effective inventory management. Nonlinear relations between demand patterns, logistics factors and external influences, however, are complicated and cannot be addressed by traditional inventory systems and shallow learning models. In order to overcome this problem, this research offers a smart inventory management system based on a Dense Nested Attention Network (DNAN) to classify states of inventory. Pre-processing of inventory information based on the sales history, logistics, and contextual information is performed by a Grid-Constrained Data Cleansing algorithm and then the most effective feature selection is done using the Electric Eel Foraging Optimization Algorithm (EeFOA). The DNAN model takes advantage of dense connectivity and hierarchical attention to effectively represent the temporal trend of demand and capture the contextual dependence of demand and stock-out risks to classify the level of demand and stock-out risk correctly. Experimental results demonstrate that the proposed DNAN-based approach achieves 99% classification accuracy, outperforming other attention-based neural models in both accuracy and computational efficiency. This work’s primary contribution is to show how dense connection and nested attention enhance decision-oriented inventory classification, providing a workable and scalable intelligent inventory control solution.