RESUMEN
Chest X-ray (CXR) imaging is widely employed by radiologists to diagnose thoracic diseases. Recently, many deep learning techniques have been proposed as computer-aided diagnostic (CAD) tools to assist radiologists in minimizing the risk of incorrect diagnosis. From an application perspective, these models have exhibited two major challenges: (1) They require large volumes of annotated data at the training stage and (2) They lack explainable factors to justify their outcomes at the prediction stage. In the present study, we developed a class activation mapping (CAM)-based ensemble model, called Ensemble-CAM, to address both of these challenges via weakly supervised learning by employing explainable AI (XAI) functions. Ensemble-CAM utilizes class labels to predict the location of disease in association with interpretable features. The proposed work leverages ensemble and transfer learning with class activation functions to achieve three objectives: (1) minimizing the dependency on strongly annotated data when locating thoracic diseases, (2) enhancing confidence in predicted outcomes by visualizing their interpretable features, and (3) optimizing cumulative performance via fusion functions. Ensemble-CAM was trained on three CXR image datasets and evaluated through qualitative and quantitative measures via heatmaps and Jaccard indices. The results reflect the enhanced performance and reliability in comparison to existing standalone and ensembled models.
RESUMEN
Rice is one of the fundamental food items that comes in many varieties with their associated benefits. It can be sub-categorized based on its visual features like texture, color, and shape. Using these features, the automatic classification of rice varieties has been studied using various machine learning approaches for marketing and industrial use. Due to the outstanding performance of deep learning, several models have been proposed to assist in vision tasks like classification and detection. Regardless of their best results on accuracy metrics, they have been observed as overly excessive for computational resources and expert supervision. To address these challenges, this paper proposes three deep learning models that offer similar performance with 10% lighter computational overhead in comparison to existing best models. Moreover, they have been trained for end-to-end flow to demonstrate minimum expert supervision for pre-processing and feature engineering sub-tasks. The results can be observed as promising for classifying rice among five varieties, namely Arborio, Basmati, Ipsala, Jasmine, and Karacadag. The process and performance of the trained models can be extended for edge and mobile devices for field-specific tasks autonomously.