Federated Learning-Powered Hybrid Model for Crop Prediction: Combining Graph Neural Networks with RNNs

FL-AGRN forecasts agricultural yields accurately while protecting privacy by combining Federated Learning (FL) with Attention-based RNN and GNN. RNN describes temporal changes, GNN captures spatial relationships, and FL allows for decentralized training. The attention mechanism improves the selection of features. A scalable precision agriculture solution is provided by the experimental results, which demonstrate exceptional accuracy and efficiency. The MATLAB application is used to execute the experimental results of the suggested strategy, showing how effective it is at increasing crop output prediction and raising overall agricultural productivity.