Machine Learning-Driven Optimization for Intrusion Detection in Smart Vehicular Networks

An essential element in the smart city vision is providing safe and secure journeys via intelligent vehicles and smart roads. Vehicular ad hoc networks (VANETs) have played a significant role in enhancing road safety where vehicles can share road information conditions. However, VANETs share the same security concerns of legacy ad hoc networks. Unlike exiting works, we consider, in this paper, detection a common attack where nodes modify safety message or drop them. Unfortunately, detecting such a type of intrusion is a challenging problem since some packets may be lost or dropped in normal VANET due to congestion without malicious action. To mitigate these concerns, this paper presents a novel scheme for minimizing the invalidity ratio of VANET packets transmissions. In order to detect unusual traffic, the proposed scheme combines evidences from current as well as past behaviour to evaluate the trustworthiness of both data and nodes. A new intrusion detection scheme is accomplished through a four phases, namely, rule-based security filter, Dempster–Shafer adder, node’s history database, and Bayesian learner. The suspicion level of each incoming data is determined based on the extent of its deviation from data reported from trustworthy nodes. Dempster–Shafer’s theory is used to combine multiple evidences and Bayesian learner is adopted to classify each event in VANET into well-behaved or misbehaving event. The proposed solution is validated through extensive simulations. The results confirm that the fusion of different evidences has a significant positive impact on the performance of the security scheme compared to other counterparts.

     

Cyberbullying Detection and Recognition with Type Determination Based on Machine Learning

Social media networks are becoming essential to our daily activities, and many issues are due to this great involvement in our lives. Cyberbullying is a social media network issue, a global crisis affecting the victims and society as a whole. It results from a misunderstanding regarding freedom of speech. In this work, we proposed a methodology for detecting such behaviors (bullying, harassment, and hate-related texts) using supervised machine learning algorithms (SVM, Naïve Bayes, Logistic regression, and random forest) and for predicting a topic associated with these text data using unsupervised natural language processing, such as latent Dirichlet allocation. In addition, we used accuracy, precision, recall, and F1 score to assess prior classifiers. Results show that the use of logistic regression, support vector machine, random forest model, and Naïve Bayes has 95%, 94.97%, 94.66%, and 93.1% accuracy, respectively.     

Age and Gender Classification Using Backpropagation andBaggingAlgorithms

Voice classification is important in creating more intelligent systems that help with student exams, identifying criminals, and security systems. The main aim of the research is to develop a system able to predicate and classify gender, age, and accent. So, a new system called Classifying Voice Gender, Age, and Accent (CVGAA) is proposed. Backpropagation and bagging algorithms are designed to improve voice recognition systems that incorporate sensory voice features such as rhythm-based features used to train the device to distinguish between the two gender categories. It has high precision compared to other algorithms used in this problem, as the adaptive backpropagation algorithm had an accuracy of 98% and the Bagging algorithm had an accuracy of 98.10% in the gender identification data. Bagging has the best accuracy among all algorithms, with 55.39% accuracy in the voice common dataset and age classification and accent accuracy in a speech accent of 78.94%.