Creating a Simulation of an Entire International Supply Chain Using Multi-Layer Perceptrons

By Adarsh Keshri and Cosminder S

Supply chains used to be predictable rivers of materials and information. Today they resemble living, global organisms--highly sensitive to minute economic shifts, political tremors, or a single late sensor reading. While volatility has become the norm, the analytical tools many companies rely on are still rooted in yesterday's assumptions. This book was written to bridge that gap. Over the past five years our research team at Cosminder Analytics has combined simulation science with modern deeplearning techniques--specifically multilayer perceptrons (MLPs)--to build fullscale digital twins for some of the world's most demanding networks. Along the way we discovered two constants: Nonlinear behavior is everywhere. Forecasting errors often hide in the interactions among dozens of small signals--fuel prices, social media sentiment, microweather, port congestion. A welltuned MLP captures those hidden links. A model is only useful when embedded in a decision cycle. Simulations that ingest realtime data, push actionable metrics, and retrain themselves outperform oneoff "whatif" studies every time.

• Format: Paperback
• Language: English
• ISBN: B0F5N5NXNJ
• ISBN13: 9798280414761
• Release Date: April 2025
• Publisher: Independently published
• Length: 200 Pages
• Dimensions: 9.0" x 0.5" x 6.0"