Tech-Driven Agronomy: Federated Learning CNN's for Aloe Vera Leaf Disease Diagnosis

The research paper discussed how integrating a federated learning framework using a Convolutional Neural Network (CNN) enables groundbreaking work in classifying aloe vera leaf diseases. The research is concerned with classifying leaf diseases into four grades, using datasets from four clients. By applying federated averaging methods, local data from each client is converted into one global unified model that preserves private information and capitalizes on collective wisdom. The result analysis offers essential findings concerning the model performance. Each client adds to the federated model uniquely, with different levels of accuracy and precision regarding disease categories. The macro, micro, and weighted averages are computed to assess the efficiency of the entire model for all customers and classes. The model for the client uj_1 has a macro average value of 78.80, a weighted average value equal to 79.25, and a micro-average of 79.36, indicating adequate initial performance. Micro, macro-averaged F1 scores of client uj_2 are 88.43 with a weighted average score at the same value. Client uj_3 refines the model's performance, scoring 90.62 in macro intermediate, 90.66 in the weighted average, and a micro according of 90. Finally, showing the exceptional performance levels is client uj_4, which attains a mean of 92.89 from macro-average, average weighted means to value at 92.0 and micro averages standing on values as high up as %. In the end, this study showed that federated learning with CNN can accurately classify aloe vera leaf diseases and establish a baseline for future research on agricultural disease management using state-of-the-art machine learning approaches.