Electronic health records (EHRs) are providing increased access to healthcare data that can be made available for advanced data analysis. This can be used by the healthcare professionals to make a more informed decision providing improved quality of care. However, due to the inherent heterogeneous and imbalanced characteristics of medical data from EHRs, data analysis task faces a big challenge. In this paper, we address the challenges of imbalanced medical data about a brain tumor diagnosis problem. Morphometric analysis of histopathological images is rapidly emerging as a valuable diagnostic tool for neuropathology. Oligodendroglioma is one type of brain tumor that has a good response to treatment provided the tumor subtype is recognized accurately. The genetic variant, 1p-/19q-, has recently been found to have high chemosensitivity, and has morphological attributes that may lend it to automated image analysis and histological processing and diagnosis. This paper aims to achieve a fast, affordable, and objective diagnosis of this genetic variant of oligodendroglioma with a novel data mining approach combining a feature selection and ensemble-based classification. In this paper, 63 instances of brain tumor with oligodendroglioma are obtained due to prevalence and incidence of the tumor variant. In order to minimize the effect of an imbalanced healthcare data set, a global optimization-based hybrid wrapper-filter feature selection with ensemble classification is applied. The experiment results show that the proposed approach outperforms the standard techniques used in brain tumor classification problem to overcome the imbalanced characteristics of medical data.

Software systems are now ubiquitous and are used every day for automation purposes in personal and enterprise applications; they are also essential to many safety-critical and mission-critical systems, e.g., air traffic control systems, autonomous cars, and SCADA systems. With the availability of massive storage capabilities, high speed Internet, and the advent of Internet of Things (IoT) devices, modern software systems are growing in both size and complexity. Maintaining a high quality of such complex systems while manually keeping the error rate at a minimum is a challenge. Therefore, automated detection of faulty components in a software system is important during software development and also post-delivery. Fault detection models usually needs to be trained on a labeled-balanced dataset with both faulty and non-faulty samples. Earlier work, e.g. Mohsin et al. (2016), showed that most real fault detection training dataset are imbalanced. Thereby, the trained model gets over-fitted and classifies faulty components as non-faulty components. The consequence of a high false negative rate is cumulative and results in generating more errors when using the model in other software systems – never seen before, which is very expensive. In this paper, we propose a software defect prediction ensemble model which considers the class imbalance problem in real software datasets. We use different oversampling techniques to build an ensemble classifier that can reduce the effect of low minority samples in the defective data. The proposed approach is verified using PROMISE software engineering dataset. The results show that our ensemble oversampling technique can more greatly reduce the false negative rate compared to the standard classification techniques and identify the faulty components more accurately resulting in a less expensive detection system (lowering the rate of non-faulty predictions of faulty modules).

Trustworthiness of an Industrial Internet of Things (IIoT) network is an important stakeholder expectation. Maintaining the trustworthiness of such a network is crucial to void loss of lives. A trustworthy IIoT system combines the security characteristics of IT trustworthiness- safety, security, privacy, reliability, and resilience. Conventional security tools and techniques are not enough to safeguard IIoT platform due to the difference in protocols, limited upgrade opportunities, mismatch in protocols and older version of the operating system used in the industrial system. In this paper, we propose to improve the trustworthiness of an IIoT network (i.e. supervisory control and data acquisition (SCADA) network) through a reliable and salable cyber-attack detection model. In particular, an ensemble-learning model based on the combination of a random subspace (RS) learning method with random tree (RT) is proposed for detecting cyber-attacks of SCADA by using the network traffics from SCADA based IIoT platform. The novelty of the proposed model is that it uses industrial protocol-based network traffic and the RS to solve the sensitivity of irrelevant features and ensemble RT to reduce the overfitting problem, thereby constructs detection engine based on industrial protocols and achieves high detection rates. The proposed model has been tested over 15 datasets of SCADA network. Experimental results reveal that the proposed model outperforms conventional detection techniques and thus improves the security and related measure of trustworthiness of IIoT platform.

Dark Web is one of the most challenging and untraceable mediums adopted by the cyber criminals, terrorists, and state-sponsored spies to fulfil their illicit motives. Cyber-crimes happening inside the Dark Web are alike the real world crimes. However, the sheer size, unpredictable ecosystem and anonymity provided by the Dark Web services are the essential confrontations to trace the criminals. To discover the potential solutions towards cyber-crimes evaluating the sailing Dark Web crime threats is a crucial step. In this paper, we will appraise the Dark Web by analysing the crimes with their consequences and enforced methods as well as future manoeuvres to lessen the crime threats. We used Systematic Literature Review (SLR) method with the aspiration to provide the direction and aspect of emerging crime threats in the Dark Web for the researchers and specialist in Cyber security field. For this SLR 65 most relevant articles from leading electronic databases were selected for data extraction and synthesis to answer our predefined research questions. The result of this systematic literature review provides (i) comprehensive knowledge on the growing crimes proceeding with Dark Web (ii) assessing the social, economic and ethical impacts of the cyber-crimes happening inside the Dark Web and (iii) analysing the challenges, established techniques and methods to locate the criminals and their drawbacks. Our study reveals that more in depth researches are required to identify criminals in the Dark Web with new prominent way, the crypto markets and Dark Web discussion forums analysis is crucial for forensic investigations, the anonymity provided by Dark Web services can be used as a weapon to catch the criminals and digital evidences should be analysed and processed in a way that follows the law enforcement to make the seizure of the criminals and shutting down the illicit sites in the Dark Web.

Digital extortion has become a major cyber risk for many organizations; small-medium enterprises (SME) to large enterprises business and individual entrepreneurs. Ransomware is a kind of malware that is the main threat to digital extortion and has caused many organizations to lose huge revenue by paying much bigger ransom demands to the cybercriminals in recent years. The explosive growth of ransomware is due to the existing large infection vector such as social engineering, email attachment, zip file download, browsing malicious site, infected search engine which are boosted dramatically by easily available cryptographic tools, Ransomware As a Service (RaaS), increased cloud storage and off-the-self ransomware toolkits. The large infection vector and available toolkits not only grew ransomware extremely, but also made them more obfuscated, encrypted and varying patterns in the new variants. This, in turn, caused the conventional supervised analysis and detection engine to fail to detect the new variants of ransomware. This paper addresses the limitations of conventional supervised detection engine and proposes semi-supervised framework to compute the inherent latent sources of the varying patterns in the new variants in an unsupervised way using deep learning approaches. The proposed framework extracts the inherent characteristics in the varying patterns from the unlabelled ransomware obtained from the wild which is scalable to accommodate upcoming malicious executables. Then the unsupervised learned model is combined with supervised classification, thus constructing an adaptive detection model. The proposed framework has been verified using real ransomware data with a dynamic analysis testbed. Our extensive experimental results and discussion demonstrate that the proposed adaptive framework can successfully identify different variants of ransomware and achieve higher performance than existing supervised approaches.

In software engineering community, defect prediction is one the active domain. For the software’s success, it is essential to reduce the software engineering and data-mining gap. Software defects prediction forecasts the source code errors before the testing phase. Methods for predicting software defects, such as clustering, statistical methods, mixed algorithms, metrics based on neural networks, black box testing, white box testing and machine learning are frequently used to explore the effect area in software. The main contribution of this research is the use of feature selection for the first time to increase the accuracy of machine learning classifiers in defects pre-diction. The objective of this study is to improve the defects prediction accuracy in five data sets of NASA namely; CM1, JM1, KC2, KC1, and PC1. These NASA data sets are open to public. In this research, the feature selection technique is use with machine-learning techniques; Random Forest, Logistic Regression, Multilayer Perceptron, Bayesian Net, Rule ZeroR, J48, Lazy IBK, Support Vector Machine, Neural Networks, and Decision Stump to achieve high defect prediction accuracy as compared to without feature selection (WOFS). The research workbench, a machine-learning tool called WEKA (Waikato Environment for Knowledge Analysis), is used to refine da-ta, preprocess data, and apply the mentioned classifiers. To assess statistical analyses, a mini tab statistical tool is used. The results of this study reveals that accuracy of defects prediction with feature selection (WFS) is improve in contrast with the accuracy of WOFS.