Dhara Solanki1, Kush Patel2 1Department of Computer Science, CSPIT, CHARUSAT, Changa, India 2Department of Information Technology, GCET, CVM University, Anand, India
Modern agriculture operates under severe information asymmetry: field-level sensor data, satellite imagery, and soil chemistry readings remain siloed across individual farms, preventing the large-scale learning that machine learning models need to generalise well. At the same time, farmers and agronomists rightly expect to understand why a model recommends a particular action before they act on it. This paper presents FedXAI-Agri, a federated learning architecture augmented with SHAP-based and LIME-based explainability modules, designed to predict crop yield and detect plant disease without transferring raw farm data to any central server. A federation of twelve simulated farm nodes contributes local gradient updates; a central aggregator applies FedAvg with differential-privacy noise clipping. The global model reaches 94.7% classification accuracy on a held-out disease dataset and reduces mean absolute error in yield prediction to 3.12 quintals per hectare, a 21.4% improvement over a centrally trained baseline trained on the same aggregate dataset. SHAP global feature rankings show soil nitrogen, relative humidity, and leaf temperature as the three most influential variables, while LIME local explanations restore full interpretability at per-prediction granularity. The framework runs at inference on a Raspberry Pi 4, making it viable for lowconnectivity edge deployments typical of rural India and sub-Saharan Africa.
Federated Learning, Explainable Artificial Intelligence, Smart Agriculture, SHAP, Differential Privacy, Crop Yield Prediction, Plant Disease Detection
Bismit Pratapsingh, Independent Researcher, Louisville, Kentucky, USA
Agentic artificial intelligence enables supply-chain systems to move beyond exception detection toward autonomous exception resolution, but fixed autonomy is unsuitable because exceptions differ in uncertainty and consequence. This paper proposes a risk-adaptive decision framework that governs AI-agent execution authority for individual supply-chain exceptions. The framework assesses six dimensions: AI confidence, exception novelty, business impact, reversibility, compliance exposure, and downstream propagation risk. Combined with non-compensatory governance guardrails, these dimensions determine one of four resolution modes: recommendation, human-approved execution, bounded autonomous resolution, or mandatory escalation. Post-execution verification preserves an auditable decision record. Analytical evaluation across eight structured procurement, inventory, order-fulfillment, and logistics scenarios demonstrates differentiated execution-authority decisions and guardrail overrides. The approach reframes autonomy as an exception-level, context-dependent governance decision for controlled and accountable deployment of agentic AI in supply-chain operations.
Agentic AI, Supply Chain Exception Management, Risk-Adaptive Autonomy, Human-AI Collaboration, Supply Chain Risk Management
Copyright © NETWORKS 2026
