This research compares various machine learning algorithms for detecting tea leaf diseases using image processing and advanced ML techniques like autoencoders, LightGBM, CatBoost, and self-organizable maps. The study demonstrates the effectiveness of ensemble methods and deep learning approaches in agricultural disease detection, with applications in precision farming and crop health monitoring.

This paper presents a novel approach to disentangle multimodal representations and quantify the contribution of each modality to the overall model performance. The work provides interpretable insights into how different modalities (text, image, audio) interact in machine learning models, with implications for improving multimodal AI systems.