Figure 2 - uploaded by Ahmad Abdel-Aziz Sharieh
Content may be subject to copyright.
Source publication
Phishing web-sites can cause the loss of thousands of dollars and leads to the damage of the brand image of organizations. Thus, automatic filtering of phishing web-site becomes a necessity. This paper presents a phishing detection technique based on Fuzzy Inference Process. The proposed phishing detection has rules for converting the input feature...
Context in source publication
Context 1
... process of assigning clusters and update the centers are repeated until converged. Figure 2 illustrates the clustering outcomes in mock dataset. For example, because samples S 2 , S 4 and S 5 are similar to each other, in term of values in their feature vector-see Figure 1, they are placed in the same cluster. ...
Similar publications
The rapidly increasing landscape of Internet-of-Thing (IoT) devices has introduced significant technical challenges for their management and security, as these IoT devices in the wild are from different device types, vendors , and product models. The discovery of IoT devices is the prerequisite to characterize, monitor, and protect these devices. H...
Citations
... Riaty et al. [76] Gamal et al. [59] Rate, F-Measure, and Matthews Correlation coefficients were considered as a criterion for the evaluation process. ...
Given the prevailing state of cybersecurity, it is reasonable to understand why cybersecurity experts are seriously considering artificial intelligence as a potential field that can aid improvements in conventional cybersecurity techniques. Various progressions in the field of technology have helped to mitigate some of the issues relating to cybersecurity. These advancements can be manifested by Big Data, Blockchain technology, Behavioral Analytics, to name but a few. The paper overviews the effects of applications of these technologies in cybersecurity. The central purpose of the paper is to review the application of AI techniques in analyzing, detecting, and fighting various cyberattacks. The effects of the implementation of conditionally classified “distributed” AI methods and conveniently classified “compact” AI methods on different cyber threats have been reviewed. Furthermore, the future scope and challenges of using such techniques in cybersecurity, are discussed. Finally, conclusions have been drawn in terms of evaluating the employment of different AI advancements in improving cybersecurity.
... But, both neural network and SVM are classifiers not easily interpretable (Otero, Freitas et al. 2013). e. [7] presented enhance detecting phishing websites based on machine learning techniques of fuzzy logic with associative rules. Hence, the accuracy phishing detection is largely dependent on feature selection. ...
Phishing is stealing users' confidential information by uploading a fake website that claims to be of another. Such a web site contains special features that aid an automatic classification of it as phishing one. However, there is no single feature that to be used to identify the phishing web-sites. Subsequently, the properties of phishing website are defined as a collection of features and then used to actively discover such sites in real-time. This research develops automatic classification of a web-site into a phishing or non-phishing one based on aggregation of a set of predetermined features related to the content of the site. A classifier is developed based on Ant-Colony optimization, known as cAnt-MinerPB. The features are combined and listed in a tree structure using document object model (DOM) representation to find the best level of detailed features that helps in capturing the phishing properties. Moreover, such hierarchical representation is used to capture strength and weakness of the classifier itself and other classifiers that are used for a comparative study. The proposed method has a fair accuracy as compared to well-known algorithms. For the tested data set, its obtained accuracy was close to that obtained by KNN and SVM. The proposed classier, for the conducted experiments, is better than other classifiers that are rule based. The cAnt-MinerPB has shown promising results compared to the well-known and well-established classification techniques.
