Vehicular Ad hoc Networks (VANETs) provide effective vehicular operation for safety as well as greener and more efficient communication of vehicles in the Dedicated Short Range Communication (DRSC). The dynamic nature of the vehicular network topology has posed many security challenges for effective communication among vehicles. Consequently, models have been applied in the literature to checkmate the security issues in the vehicular networks. Existing models lack flexibility and sufficient functionality in capturing the dynamic behaviors of malicious nodes in the highly volatile vehicular communication systems. Given that existing models have failed to meet up with the challenges involved in vehicular network topology, it has become imperative to adopt complementary measures to tackle the security issues in the system. The approach of trust model with respect to Machine/Deep Learning (ML/DL) is proposed in the paper due to the gap in the area of network security by the existing models. The proposed model is to provide a data-driven approach in solving the security challenges in dynamic networks. This model goes beyond the existing works conceptually by modeling trust as a classification process and the extraction of relevant features using a hybrid model like Bayesian Neural Network that combines deep learning with probabilistic modeling for intelligent decision and effective generalization in trust computation of honest and dishonest nodes in the network.

Connected and automated vehicles (CAVs) as a part of Intelligent Transportation Systems (ITS) are projected to revolutionise the transportation industry, primarily by allowing real-time and seamless information exchange of vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I). However, these connectivity and automation are expected to offer vast numbers of benefits, new challenges in terms of safety, security and privacy also emerge. CAVs continue to rely heavily on their sensor readings, the input obtained from other vehicles and the road side units to inspect roadways. Consequently, anomalous reading of sensors triggered by malicious cyber attacks may lead to fatal consequences. Hence, like all other safety-critical applications, in CAVs also, reliable and secure information dissemination is of utmost importance. As a result, real time detection of anomaly along with identifying the source is a pre-requisite for mass deployment of CAVs. Motivated by this safety concerns in CAVs, we develop an efficient anomaly detection method through the combination of Bayesian deep learning (BDL) with discrete wavelet transform (DWT) to improve the safety and security in CAVs. In particular, DWT is used to smooth sensor reading of a CAV and then feed the data to a BDL module for analysis of the detection and identification of anomalous sensor behavior/data points caused by either malicious cyber attacks or faulty vehicle sensors. Our numerical experiments show that the proposed method demonstrates significant improvement in detection anomalies in terms of accuracy, sensitivity, precision, and F1-score evaluation metrics. For these metrics, the proposed method shows an average performance gain of $7.95\%$, $9\%$, $8.77\%$ and $7.33\%$, respectively when compared with Convolutional Neural Network (CNN-1D), and when compared with BDL, the corresponding numbers are $5\%$, $7.9\%$, $7.54\%$ and $4.1\%$ respectively.

The Machine Type Communication Devices (MTCDs) are usually based on Internet Protocol (IP), which can cause billions of connected objects to be part of the Internet. The enormous amount of data coming from these devices are quite heterogeneous in nature, which can lead to security issues, such as injection attacks, ballot stuffing, and bad mouthing. Consequently, this work considers machine learning trust evaluation as an effective and accurate option for solving the issues associate with security threats. In this paper, a comparative analysis is carried out with five different machine learning approaches: Naive Bayes (NB), Decision Tree (DT), Linear and Radial Support Vector Machine (SVM), KNearest Neighbor (KNN), and Random Forest (RF). As a critical element of the research, the recommendations consider different Machine-to-Machine (M2M) communication nodes with regard to their ability to identify malicious and honest information. To validate the performances of these models, two trust computation measures were used: Receiver Operating Characteristics (ROCs), Precision and Recall. The malicious data was formulated in Matlab. A scenario was created where 50% of the information were modified to be malicious. The malicious nodes were varied in the ranges of 10%, 20%, 30%, 40%, and the results were carefully analyzed.

Security challenges present in Machine-to-Machine Communication (M2M-C) and big data paradigm are fundamentally different from conventional network security challenges. In M2M-C paradigms, “Trust” is a vital constituent of security solutions that address security threats and for such solutions,it is important to quantify and evaluate the amount of trust in the information and its source. In this work, we focus on Machine Learning (ML) Based Trust (MLBT) evaluation model for detecting malicious activities in a vehicular Based M2M-C (VBM2M-C) network. In particular, we present an Entropy Based Feature Engineering (EBFE) coupled Extreme Gradient Boosting (XGBoost) model which is optimized with Binary Particle Swarm optimization technique. Based on three performance metrics, i.e., Accuracy Rate (AR), True Positive Rate (TPR), False Positive Rate (FPR), the effectiveness of the proposed method is evaluated in comparison to the state-of-the-art ensemble models, such as XGBoost and Random Forest. The simulation results demonstrates the superiority of the proposed model with approximately 10% improvement in accuracy, TPR and FPR, with reference to the attacker density of 30% compared with the start-of-the-art algorithms.

Malicious attacks on Connected and Automated Vehicle (CAV) systems can severely impact transportation safety and efficiency. They can manipulate sensor readings, disrupt service, or falsify environmental perceptions, leading to incorrect decisions and potential catastrophes. Detecting such anomalies is complex due to challenges such as false positives and negatives, sensor interference, resource constraints, and diverse types of anomalies. Existing traditional methods, including statistical approaches, signature-based detection, behavioral analysis, and network intrusion detection systems, often suffer from high false positive rates, sensitivity to noise, reliance on thresholds, and inability to detect evolving or subtle patterns. Recent machine learning and hybrid methods improve on traditional techniques but struggle with frequency bias and time scale selection, limiting real-time applications in CAVs networks. To address these challenges, we propose ET-DRF-GA, a novel method that combines an ensemble-based exploration of Temporal Convolutional Networks (TCN), Differential Receptive Fields (DRF), and Genetic Algorithms (GA). This approach offers dynamic parameter adjustment via genetic algorithms, enhanced robustness and stability through ensemble-based model selection, and optimized sensitivity to different anomaly types using DRF. Experiments on benchmark datasets, Safety Pilot Model Deployment (SPMD) and Vehicular Reference Misbehavior (VeReMi), demonstrate the method’s robustness and effectiveness in anomaly detection for CAV systems, paving the way for future research in this critical domain.

Heart failure (HF) prediction models using machine learning (ML) must achieve a balance between performance, fairness, and real-world clinical utility. This paper assesses the potential of ML and DL models in the context of heterogeneous databases (UCI, MIMIC) and aims to derive applicable schemes for equitable deployment in healthcare. Although the Transformer models depicted notable AUC-ROC in the UCI data (0.986), they suffered a considerable performance degradation in the MIMIC data, suggesting their generalizability issue. Interestingly, we found contrasting gender preferences: SVM better detected females (AUC-ROC = 0.869 vs 0.796 for males), while XGBoost favored males. Decision curve analysis (DCA) demonstrated that a greater AUC-ROC was not necessarily preferred. For the 0.5 threshold, the net benefit of SVM (0.050) was higher than that of other models. SHAP analysis confirmed sex, ejection fraction, and NT-proBNP as the main predictors, but demonstrated inconsistent feature relations between datasets. We argue that (1) gender-specific threshold optimization, (2)ensemble methods to counteract bias, and (3)robust external validation are necessary and provide a possible path for clinically deployable, fairness-aware HF prediction tools.