Spam Email Detection Using Deep Learning Techniques

Machine learning algorithms are widely utilized in cybersecurity. However, recent studies show that machine learning algorithms are vulnerable to adversarial examples. This poses new threats to the security-critical applications in cybersecurity. Currently, there is still a short of study on adversarial examples in the domain of cybersecurity. In this paper, we propose a new method known as the brute-force attack method to better evaluate the robustness of the machine learning classifiers in cybersecurity against adversarial examples. The proposed method, which works in a black-box way and covers some shortages of the existing adversarial attack methods based on generative adversarial networks, is simple to implement and only needs the output of the target classifiers to generate adversarial examples. To have a comprehensive evaluation of the attack performance of the proposed method, we use our method to generate adversarial examples against the common machine learning based security systems in cybersecurity including host intrusion detection systems, Android malware detection systems, and network intrusion detection systems. We compare the attack performance of the proposed method against these security systems with that of state-of-the-art adversarial attack methods based on generative adversarial networks. The preliminary experimental results show that the proposed method, which is more efficient in computation and outperforms the state-of-the-art attack methods based on generative adversarial networks, can be used to evaluate the robustness of various machine learning based systems in cybersecurity against adversarial examples.

/pdf/a-brute-force-black-box-method-to-attack-machine-learning-40y9yilzd2.pdf

A Brute-Force Black-Box Method to Attack Machine Learning-Based Systems in Cybersecurity

The paper elaborates on how text analysis influences classification—a key part of the spam-filtering process. The authors propose a multistage meta-algorithm for checking classifier performance. As a result, the algorithm allows for the fast selection of the best-performing classifiers as well as for the analysis of higher-dimensionality data. The last aspect is especially important when analyzing large datasets. The approach of cross-validation between different datasets for supervised learning is applied in the meta-algorithm. Three machine-learning methods allowing a user to classify e-mails as desirable (ham) or potentially harmful (spam) messages were compared in the paper to illustrate the operation of the meta-algorithm. The used methods are simple, but as the results showed, they are powerful enough. We use the following classifiers: k-nearest neighbours (k-NNs), support vector machines (SVM), and the naive Bayes classifier (NB). The conducted research gave us the conclusion that multinomial naive Bayes classifier can be an excellent weapon in the fight against the constantly increasing amount of spam messages. It was also confirmed that the proposed solution gives very accurate results.

A Method for Fast Selection of Machine-Learning Classifiers for Spam Filtering

Abstract Communication via email has expanded dramatically in recent decades due to its cost-effectiveness, convenience, speed, and utility for a variety of contexts, including social, scientific, cultural, political, authentication, and advertising applications. Spam is an email sent to a large number of individuals or organizations without the recipient's desire or request. It is increasingly becoming a harmful part of email traffic and can negatively affect the usability of email systems. Such emails consume network bandwidth as well as storage space, causing email systems to slow down, wasting time and effort scanning and eliminating enormous amounts of useless information. Spam is also used for distributing offensive and harmful content on the Internet. The objective of the current study was to develop a new method for email spam detection with high accuracy and a low error rate. There are several methods to recognize, detect, filter, categorize, and delete spam emails, and almost the majority of the proposed methods have some extent of error rate. None of the spam detection techniques, despite the optimizations performed, have been effective alone. A step in text mining and message classification is feature selection, and one of the best approaches for feature selection is the use of metaheuristic algorithms. This article introduces a new method for detecting spam using the Horse herd metaheuristic Optimization Algorithm (HOA). First, the continuous HOA was transformed into a discrete algorithm. The inputs of the resulting algorithm then became opposition-based and then converted to multiobjective. Finally, it was used for spam detection, which is a discrete and multiobjective problem. The evaluation results indicate that the proposed method performs better compared to other methods such as K-nearest neighbours-grey wolf optimisation, K-nearest neighbours, multilayer perceptron, support vector machine, and Naive Bayesian. The results show that the new multiobjective opposition-based binary horse herd optimizer, running on the UCI data set, has been more successful in the average selection size and classification accuracy compared with other standard metaheuristic methods. According to the findings, the proposed algorithm is substantially more accurate in detecting spam emails in the data set in comparison with other similar algorithms, and it shows lower computational complexity. 

/pdf/a-novel-approach-for-spam-detection-using-horse-herd-2s686d3t.pdf

A novel approach for spam detection using horse herd optimization algorithm

Although the problem of spam classification seems to be solved, there are still vulnerabilities in the current spam filters that could be easily exploited. We present one such vulnerability, in which one could replace some characters with corresponding characters from a different alphabet. These characters are visually similar, yet have a different Unicode encoding. With this approach spammers can create messages that bypass existing spam filters. Moreover, we show that this approach can be used to avoid plagiarism detection, and in other applications that use natural language processing for automatic analysis of text documents.

Visual Spoofing in content based spam detection

$E$ -mail has become an integral part of day-to-day life. It has become one of the important and convenient forms of communication for both professional as well as personal use. This type of communication saves a lot of time of human beings. Emails can be considered as either spam or ham. Spam has become one of the major problems to internet users. It is very important to get rid of spam mails. Generally, users do not have much control over the accessing of mails and have to accept whatever that comes in to their inbox. Many spam filtering techniques has been introduced to overcome the problems caused by spam. In this paper, a comparison between different machine learning classifiers used for spam classification is presented.

Spam Filtering: A Comparison Between Different Machine Learning Classifiers

Emails are inevitable in this modern era. It has become an effective tool in the communication field. Because of its easiness the number of users are increasing day-by-day. With the increased number of email users, the number of spam mails have also increased. Spam can cause great loss to users. Many spam filtering techniques have been introduced to distinguish between ham and spam. A mail that appear as spam may appear as ham to another user and vice versa. That is, it depends on the personal preference of user. Therefore an ontology based personalized mail access is necessary. This paper compares the ontology based spam filtering and conventional spam filtering.

Nasreen M Shajideen

Papers

Spam Filtering: A Comparison Between Different Machine Learning Classifiers

Conventional and Ontology Based Spam Filtering