Loading content...
Plant diseases pose a major threat to global agricultural productivity and crop quality, necessitating the development of accurate, reliable, and interpretable prediction systems. The proposed explainable hybrid ensemble learning framework is justified by its ability to address the limitations of traditional single-model approaches, which often struggle with generalization and robustness across diverse crop types and environmental conditions. By integrating multiple machine learning and deep learning models such as Support Vector Machine (SVM), K-Nearest Neighbors (KNN), Random Forest (RF), and Convolutional Neural Network (CNN), the framework leverages the strengths of each technique to enhance overall predictive performance and reduce model bias and variance. The inclusion of explainability methods like SHAP and LIME further strengthens the approach by providing transparency in decision-making, enabling farmers and agronomists to understand the key factors influencing disease predictions. The high performance metrics achieved, including accuracy of 97.9% and AUC of 99.1%, demonstrate the effectiveness of the model in accurately identifying plant diseases while minimizing misclassification. This combination of high accuracy and interpretability supports early disease detection, facilitates informed agricultural decisions, and promotes sustainable farming practices. Consequently, the framework significantly contributes to the advancement of precision agriculture by offering a scalable, trustworthy, and data-driven solution for improving crop health and productivity.
Plant diseases remain a critical challenge to agriculture, significantly reducing crop yield and quality while threatening global food security and economic stability. Traditional disease detection methods rely heavily on expert knowledge and manual field inspections, which are often labour-intensive, time-consuming, and susceptible to human error. These limitations highlight the urgent need for advanced computational approaches capable of accurately predicting plant diseases at early stages, thereby enabling timely intervention and improved crop management. Recent advancements in machine learning (ML) and deep learning (DL), particularly in image-based plant disease diagnosis, have demonstrated strong potential by extracting complex patterns from visible symptoms such as leaf discoloration, lesions, and structural changes. However, despite their success, most existing models face practical challenges in real-world agricultural environments, including limited interpretability, poor generalisation across different crop types, and lack of robustness under varying environmental conditions.
To overcome these challenges, hybrid ensemble learning techniques have emerged as an effective solution. By combining multiple predictive models, ensemble approaches enhance model stability, reduce overfitting, and improve overall accuracy. These hybrid systems integrate the strengths of diverse classifiers such as decision trees, support vector machines, convolutional neural networks, and recurrent neural networks, enabling them to capture both spatial and temporal characteristics of plant disease progression. Furthermore, the integration of explainable artificial intelligence (XAI) techniques has become increasingly important, as it ensures transparency in model predictions. Methods such as SHAP and LIME allow agronomists and farmers to understand the key factors influencing disease detection, thereby increasing trust and facilitating adoption. In the agricultural domain, where decision-making directly impacts productivity and sustainability, interpretability and reliability are essential, making explainable hybrid ensemble frameworks a highly justified and impactful approach.