Search

Tuna fish is a popular food because of its nutritional value and taste. Demand for various species of tuna increases over time, necessitating the development of a system to sort tuna fish into distinct species in export sectors in order to accelerate the process. The work proposes an automated tuna classification system based on split octonion network. The images are initially preprocessed and divided into region images. Each region image is applied to a split octonion network with eleven layers. In addition, a split octonion channel attention module is presented, which is fed to the last two convolutional layers. The features from the three octonion networks are fused and applied to a series of dense layers. In the last layer, a softmax classifier is utilized for final classification. Results show that the proposed region-based split octonion network with attention module gives an accuracy of 98.01% on tuna database. The region-based tuna classification model is fine-tuned for the categorization of six species from QUT-FishBase dataset and Fish-Pak dataset. The system shows accuracies of 97.83% and 98.17% on QUT-FishBase and Fish-Pak datasets, respectively. The proposed methodology is also compared with existing approaches using a variety of evaluation criteria. [ABSTRACT FROM AUTHOR]

Disease detection software that works automatically in healthcare domain refers to the proactive or reactive use of computerised data systems for diagnosis of diseases. Medical knowledge base, data processing, and data analytics are the three major components of the system. The procedures of data processing and data analytics are crucial. Data mining (DM) techniques were used to process these processes. DM is a tool for finding patterns in massive amounts of data and retrieving knowledge. Clinical and diagnostic evidence has created a slew of reliable timely detection services and other health-related technology in the DM and healthcare industries. Artificial Intelligence (AI) in Machine Learning (ML) includes classification and predictive analytics. Identifying key characteristics and developing a classification model to determine whether the cases are disease or not is a difficult task. Feature selection (FS) refers to the process of reducing the quantity of input features when developing a predictive model. Reducing the number of input features is desirable because it cuts the computational cost of modelling while also improving the model's performance in some cases. Instead of using a single feature selection, Optimal Ensemble Feature Selection (OEFS) solves a feature selection problem by integrating numerous feature selections. The OEFS method works by integrating the outputs of different single feature selection models like Divergence Weight Elephant Herding Optimization (DWEHO), Divergence Weight ButterFly Optimization Algorithm (DWBFO), and Differential Evolution (DE). By merging different subsets of features, Weighted Majority Voting (WMV) is used in finding the optimal feature subset. Classification model using Kernel Weight Convolutional Neural Network (KWCNN) classification is proposed. The convolution operation is a mathematical linear action across matrices that gives it its name. In terms of medical disease diagnosis, the proposed KWCNN classification performs quite well. To determine the performance of all classification algorithms, evaluation criteria such as sensitivity, specificity, f-measure, and accuracy were measured using a confusion matrix. [ABSTRACT FROM AUTHOR]

Digital multimedia images are treated as evidences in the court of law around the globe, but manipulation and tampering of photographs have become common due to the use of efficient image editing software. The most common counterfeit technique is the copy-move forgery (CMF) where a part of an original image is cloned on to the same image once or multiple times so that it may not be detected easily by human eyes. In this paper, an efficient as well as a robust algorithm for the detection of CMF is proposed. In the proposed algorithm, the suspicious image undergoes block tiling and features are extracted from each block by applying singular value decomposition (SVD) and discrete orthonormal Stockwell transform (DOST). Feature vectors obtained are lexicographically sorted and similarities of image blocks are identified by two threshold values. Simulated results show that the proposed algorithm is more invariant and robust to post-processing operations like Gaussian blurring, Gaussian white noise contamination, rotation and JPEG compression in comparison with other state-of-the-art techniques. The method offers high detection accuracy rate (DAR) with less computational complexity, even for detecting multiple copy-move forgeries. [ABSTRACT FROM AUTHOR]

Electronic auscultation is an efficient technique to evaluate the condition of respiratory system using lung sounds. As lung sound signals are non-stationary, the conventional method of frequency analysis is not highly successful in diagnostic classification. This paper deals with a novel method of analysis of lung sound signals using wavelet transform, and classification using artificial neural network (ANN). Lung sound signals were decomposed into the frequency subbands using wavelet transform and a set of statistical features was extracted from the subbands to represent the distribution of wavelet coefficients. An ANN based system, trained using the resilient backpropagation algorithm, was implemented to classify the lung sounds to one of the six categories: normal, wheeze, crackle, squawk, stridor, or rhonchus. [Copyright &y& Elsevier]