Forest fires are a serious ecological concern, and smoke is an early warning indicator. Early smoke images barely capture a tiny portion of the total smoke. Because of the irregular nature of smoke’s dispersion and the dynamic nature of the surrounding environment, smoke identification is complicated by minor pixel-based traits. This study presents a new framework that decreases the sensitivity of various YOLO detection models. Additionally, we compare the detection performance and speed of different YOLO models such as YOLOv3, YOLOv5, and YOLOv7 with prior ones such as Fast R-CNN and Faster R-CNN. Moreover, we follow the use of a collected dataset that describes three distinct detection areas, namely close, medium, and far distance, to identify the detection model’s ability to recognize smoke targets correctly. Our model outperforms the gold-standard detection method on a multi-oriented dataset for detecting forest smoke by an mAP accuracy of 96.8% at an IoU of 0.5 using YOLOv5x. Additionally, the findings of the study show an extensive improvement in detection accuracy using several data-augmentation techniques. Moreover, YOLOv7 outperforms YOLOv3 with an mAP accuracy of 95%, compared to 94.8% using an SGD optimizer. Extensive research shows that the suggested method achieves significantly better results than the most advanced object-detection algorithms when used on smoke datasets from wildfires, while maintaining a satisfactory performance level in challenging environmental conditions.

Agile software development does not include elicitation and management of non-functional requirements (NFRs). Several techniques have been proposed by researchers to elicit, manage and prioritize NFRs. The major issue with such techniques is that they involve customer or users to elicit and prioritize NFRs, which becomes problematic because they do not have exposure of NFRs. They may only be familiar with very few common NFRs such as performance, efficiency and security etc. There could be several other NFRs such as confidentiality, scalability, usability, maintainability, portability, reliability and many others which may not be known to the stakeholders. Consequently, some relevant NFRs may be missed and some wrong NFRS may be implemented in the final product which can degrade its overall quality. Secondly, it is very time consuming to elicit and prioritize NFRs from the stakeholders who have very low exposure. It is especially critical in the case of agile software development where product is developed and delivered at a very rapid pace. To overcome the challenge, this study proposed a model to prioritize NFRs for agile software development to produce quality software products. Thus, firstly, an extensive literature review was conducted to identify the NFRs which were further ranked through a survey carried out in agile software companies. Moreover, the NFRs were analyzed and prioritized using multi criteria decision making analysis technique called analytical hierarchical technique (AHP). The evaluation results showed that the NFRs prioritization model reduce the overhead of meetings and discussion required to gather and identify the NFRs common to agile products.

Developing an effective classification model in the medical field is challenging due to limited datasets. To address this issue, this study proposes using a generative adversarial network (GAN) as a data-augmentation technique. The research aims to enhance the classifier’s generalization performance, stability, and precision through the generation of synthetic data that closely resemble real data. We employed feature selection and applied five classification algorithms to thirteen benchmark medical datasets, augmented using the least-square GAN (LS-GAN). Evaluation of the generated samples using different ratios of augmented data showed that the support vector machine model outperforms other methods with larger samples. The proposed data augmentation approach using a GAN presents a promising solution for enhancing the performance of classification models in the healthcare field.

Hepatitis C is a significant public health concern, resulting in substantial morbidity and mortality worldwide. Early diagnosis and effective treatment are essential to prevent the disease’s progression to chronic liver disease. Machine learning algorithms have been increasingly used to develop predictive models for various diseases, including hepatitis C. This study aims to evaluate the performance of several machine learning algorithms in diagnosing chronic liver disease, with a specific focus on hepatitis C, to improve the cost-effectiveness and efficiency of the diagnostic process. We collected a comprehensive dataset of 1801 patient records, each with 12 distinct features, from Jordan University Hospital. To assess the robustness and dependability of our proposed framework, we conducted two research scenarios, one with feature selection and one without. We also employed the Sequential Forward Selection (SFS) method to identify the most relevant features that can enhance the model’s accuracy. Moreover, we investigated the effect of the synthetic minority oversampling technique (SMOTE) on the accuracy of the model’s predictions. Our findings indicate that all machine learning models achieved an average accuracy of 83% when applied to the dataset. Furthermore, the use of SMOTE did not significantly affect the accuracy of the model’s predictions. Despite the increasing use of machine learning models in medical diagnosis, there is a growing concern about their interpretability. As such, we addressed this issue by utilizing the Shapley Additive Explanations (SHAP) method to explain the predictions of our machine learning model, which was specifically developed for hepatitis C prediction in Jordan. This work provides a comprehensive evaluation of various machine learning algorithms in diagnosing chronic liver disease, with a particular emphasis on hepatitis C. The results provide valuable insights into the cost-effectiveness and efficiency of the diagnostic process and highlight the importance of interpretability in medical diagnosis.

Machine learning has become ubiquitous across industries for its ability to uncover in- sights from data. This research explores the application of machine learning for identifying phishing websites. The efficiency of different algorithms at classifying malicious sites is evaluated and contrasted. By exposing the risks of phishing, the study aims to develop reliable systems for fake website detection. The results showcase machine learning's capabilities for augmented cybersecurity through automated threat intelligence. Phishing employs social engineering techniques to disguise malicious links as trusted entities, tricking victims into revealing sensitive information. This work investigates phishing detection leveraging curated lists and machine learning for adaptive defense.