An Intelligent Multi-Task Supply Chain Model Based on Bio-Inspired Networks

Area of Science:

• Artificial Intelligence
• Supply Chain Management
• Computational Neuroscience

Background:

• Deep neural networks are inspired by biological systems, with convolutional neural networks (CNNs) mimicking visual cortex processing and graph neural networks (GNNs) simulating neuronal communication.
• Intelligent supply chains require agile, resilient, and sustainable systems, with network sustainability being crucial for overall performance.

Purpose of the Study:

• To propose a novel bio-inspired deep ensemble network, the Chebyshev ensemble graph network (Ch-EGN), for creating an intelligent supply chain model.
• To enhance supply chain sustainability, improve risk administration, identify hidden risks, and increase transparency.
• To evaluate the Ch-EGN's functionality on real-world supply chain datasets.

Main Methods:

• Developed a hybrid learning approach using a novel deep ensemble network, the bio-inspired Chebyshev ensemble graph network (Ch-EGN).
• Leveraged principles from both CNNs and GNNs, inspired by biological neural processing.
• Assessed the network's performance on the SupplyGraph and DataCo databases for various supply chain tasks.

Main Results:

• Achieved an average accuracy of 98.95% for automatic delivery status prediction.
• Demonstrated significant improvements in risk administration, supply chain sustainability, and transparency.
• Validated the Ch-EGN's efficiency in multi-class categorization scenarios for intelligent supply chains.

Conclusions:

• The proposed bio-inspired Chebyshev ensemble graph network (Ch-EGN) is an effective hybrid learning model for intelligent supply chains.
• The Ch-EGN significantly enhances supply chain sustainability and risk management capabilities.
• The approach offers a promising direction for developing more agile, resilient, and transparent supply chain systems.