Search

In the field of medical imaging, the increasing demand for advanced computer-aided diagnosis systems is crucial in radiography. Accurate identification of various diseases, such as COVID-19, pneumonia, tuberculosis, and pulmonary lung nodules, holds vital significance. Despite substantial progress in the medical field, a persistent research gap necessitates the development of models that excel in precision and provide transparency in decision-making processes. In order to address this issue, this work introduces an approach that utilizes transfer learning through the EfficientNet-B4 architecture, leveraging a pre-trained model to enhance the classification performance on a comprehensive dataset of lung X-rays. The integration of explainable artificial intelligence (XAI), specifically emphasizing Grad-CAM, contributes to model interpretability by providing insights into the neural network’s decision-making process, elucidating the salient features and activation regions influencing multi-disease classifications. The result is a robust multi-disease classification system achieving an impressive 96% accuracy, accompanied by visualizations highlighting critical regions in X-ray images. This investigation not only advances the progression of computer-aided diagnosis systems but also sets a pioneering benchmark for the development of dependable and transparent diagnostic models for lung disease identification.

Abstract Grey Wolf optimization (GWO) is a newly developed stochastic meta-heuristic technique motivated by nature. It shows potential in diverse optimization challenges. It replicates grey wolf hunting behaviour and social hierarchy, exploring the solution space similar to their natural process. The algorithm efficiently explores and converges to the optimal solution. However, a drawback of the standard GWO is its limited exploitation capability due to its exploration-focused iterations. This may hinder finding the optimal solution nearby, leading to lower local convergence rates and degraded solution quality. To address this, the GWO-Employed-Onlooker model suggests incorporating the onlooker and scout bee operators from the artificial bee colony algorithm (ABC) during the position-changing stage of the grey wolves. This enhances exploitation capability, resulting in improved local convergence rates and better solution quality. The proposed method’s performance is evaluated on various optimization functions and compared their convergence rate to standard GWO, Genetic Algorithm (GA), Firefly Algorithm (FA), ABC, and Ant Colony Optimization (ACO) techniques. The results demonstrate that the proposed strategy GWO-Employed-Onlooker is better, indicating that it is valuable in solving optimization problems.

A network intrusion detection system is critical for cyber security against illegitimate attacks. In terms of feature perspectives, network traffic may include a variety of elements such as attack reference, attack type, a sub-category of attack, host information, malicious scripts, etc. In terms of network perspectives, network traffic may contain an imbalanced number of harmful attacks when compared to normal traffic. It is challenging to identify a specific attack due to complex features and data imbalance issues. To address these issues, this paper proposes an Intrusion Detection System using transformer-based transfer learning for Imbalanced Network Traffic (IDS-INT). IDS-INT uses transformer-based transfer learning to learn feature interactions in both network feature representation and imbalanced data. First, detailed information about each type of attack is gathered from network interaction descriptions, which include network nodes, attack type, reference, host information, etc. Second, the transformer-based transfer learning approach is developed to learn detailed feature representation using their semantic anchors. Third, the Synthetic Minority Oversampling Technique (SMOTE) is implemented to balance abnormal traffic and detect minority attacks. Fourth, the Convolution Neural Network (CNN) model is designed to extract deep features from the balanced network traffic. Finally, the hybrid approach of the CNN-Long Short-Term Memory (CNN-LSTM) model is developed to detect different types of attacks from the deep features. Detailed experiments are conducted to test the proposed approach using three standard datasets, i.e., UNSW-NB15, CIC-IDS2017, and NSL-KDD. An explainable AI approach is implemented to interpret the proposed method and develop a trustable model.