The quality of vegetables and fruits are judged by their visual features. Misclassification of fruits and vegetables lead to a financial loss. To prevent the loss, superstores need to classify fruits and vegetables in terms of size, color and shape. To improve the accuracy of models for classifying fruits and vegetables, researchers have introduced various CNN architectures like VGG16, AlexNet, DenseNet-121 and different attention-based variants by compromising models' complexity leading to high computational cost and overlooking feature's contribution for interpretable predictions. Additionally, the major drawback in most of the existing works is the utilization of the limited dataset and number of classes for fruit and vegetable classification task instead of the benchmark dataset like Fruit-360 which comprises of 141 distinct fruit and vegetable classes. In this study, an explainable artificial intelligence (XAI) driven enhanced attention-CNN (EA-CNN) is proposed for accomplishing the fruit and vegetable classification task accurately and efficiently. The proposed EA-CNN model (a) detects the class of fruits and vegetable accurately through visual features by manipulating undiscover customized pooling technique and enhanced attention feature extraction mechanism, (b) classifies the fruits and vegetables efficiently through a simplified architecture, (c) provides interpretable predictions through XAI approach. By utilizing entire fruit-360 database, this research noteworthily enhances the model's capability to classify wide range of fruits and vegetables, thus providing an effective and reliable solution for practical applications. The proposed study outclasses baseline models with regard to accuracy and computational cost on fruit-360 benchmark dataset. EA-CNN achieves better accuracy of 98.1 % in smaller number of iterations as compared to benchmark models. Moreover, to further prove the efficiency, generalization ability, robustness, scalability and adaptability of this research, the architecture of proposed EA-CNN model is validated on another real-world dataset named as ‘Fruit Recognition’. The experimental outcomes display that the proposed EA-CNN model produces substantial performance on Fruit-Recognition dataset as well by attaining a generalized accuracy of 96 %.
Alzheimer's is one of the brain syndromes that steadily affects the brain memory. The early stage of Alzheimer's disease (AD) is referred to as mild cognitive impairment (MCI), and the growth of Alzheimer's is not certain in patients with MCI. The premature detection of Alzheimer's is crucial for maintaining healthy brain function and avoiding memory loss. Different multi-neural network architectures have been proposed by researchers for efficient and accurate AD detection. The absence of improved feature extraction mechanisms and unexplored efficient optimizers in complex benchmark architectures lead to an inefficient and inaccurate AD classification. Moreover, the standard convolutional neural network (CNN)-based architectures for Alzheimer's diagnosis lack interpretability in their predictions. An interpretable, simplified, yet effective deep learning model is required for the accurate classification of AD. In this study, a generalized fractional order-based CNN classifier with explainable artificial intelligence (XAI) capabilities is proposed for accurate, efficient, and interpretable classification of AD diagnosis. The proposed study (a) classifies AD accurately by incorporating unexplored pooling technique with enhanced feature extraction mechanism, (b) provides fractional order-based optimization approach for adaptive learning and fast convergence speed, and (c) suggests an interpretable method for proving the transparency of the model. The proposed model outperforms complex benchmark architectures with regard to accuracy using standard ADNI dataset. The proposed fractional order-based CNN classifier achieves an improved accuracy of 99 % as compared to the state-of-the-art models.
21 December 2024 in Computers in Biology and Medicine
Deep learning (DL) has had an incredible influence on many different scientific areas over the past couple of decades. Particularly in the field of healthcare, DL strategies were able to outclass other existing methodologies in image processing. The rapid expansion of the monkeypox endemic to over 40 nations apart from Africa has prompted serious worries in the realm of public health. Given that monkeypox can have symptoms that are akin to both chickenpox and measles, early detection can be difficult. Fortunately, due to the developments in artificial intelligence approaches, it can be implemented to promptly and accurately identify monkeypox disease using visual data information. Many DL driven techniques have already been exploited in the literature for skin related issues, which have provided accurate results to some extent. These models were dependent on extensive computational and time resources due to which the real-time applicability is difficult. Rather of building and training CNNs from scratch, this study uses transfer learning (TL) technique to fine-tune pre-trained networks, particularly exploiting various versions of ConvNeXt, by substituting last layer with additional task specific ones. A number of pre-processing and data augmentation methods have also been assessed and adjusted with regard to computing time and performance. The proposed study performs the binary and multi class monkeypox disease classification task. Promising accurate results of 99.9 % on the benchmark MSLD (binary class) dataset and 94 % on the MSLD v2.0 (multi-class) dataset is obtained by fine-tuned TL-based ConvNeXtSmall and ConvNeXtBase architecture with Adafactor optimization technique, demonstrating the practicality of the suggested framework as a substitute for the current ones. The proposed model is assessed through both standard train-test split and k-fold cross validation techniques. Furthermore, performance of models is also assessed on several other metrics including recall, F1 score, precision and multiple statistical tests incorporated with explainable AI methods for better interpretability of results. The concerns regarding the real-time applicability are tackled by utilizing the less time consuming and computationally efficient networks through the exploitation of transfer learning capabilities. Moreover, the explainable findings of the proposed study will be highly valuable for the healthcare professionals to understand the decisive behavior of the model and make informed clinical decisions.