Smart agriculture integrates modern digital technologies to enhance farming productivity and sustainability. This paper presents an updated Intelligent Machine Learning-based Decision Support System (IML-DSS) that incorporates Edge AI and Federated Learning for real-time agricultural decision-making. The system predicts crop yield, detects plant diseases, and optimizes resource utilization using multi-source data such as IoT sensors, satellite imagery, and climate databases. Advanced models including ensemble learning, transformer-based architectures, and explainable AI are utilized to process heterogeneous agricultural data. Edge-cloud integration enables low-latency processing and scalability. Experimental validation using recent datasets shows improved prediction accuracy, faster decision-making, reduced environmental impact, and enhanced farmer trust through interpretability. The proposed system contributes significantly to sustainable and data-driven agriculture.