The rapid adoption of deep learning in sensitive domains has brought tremendous benefits. However, this widespread adoption has also given rise to serious vulnerabilities, particularly model inversion (MI) attacks, posing a significant threat to the privacy and integrity of personal data. The increasing prevalence of these attacks in applications such as biometrics, healthcare, and finance has created an urgent need to understand their mechanisms, impacts, and defense methods. This survey aims to fill the gap in the literature by providing a structured and in-depth review of MI attacks and defense strategies. Our contributions include a systematic taxonomy of MI attacks, extensive research on attack techniques and defense mechanisms, and a discussion about the challenges and future research directions in this evolving field. By exploring the technical and ethical implications of MI attacks, this survey aims to offer insights into the impact of AI-powered systems on privacy, security, and trust. In conjunction with this survey, we have developed a comprehensive repository to support research on MI attacks and defenses. The repository includes state-of-the-art research papers, datasets, evaluation metrics, and other resources to meet the needs of both novice and experienced researchers interested in MI attacks and defenses, as well as the broader field of AI security and privacy. The repository will be continuously maintained to ensure its relevance and utility. It is accessible at https://github.com/overgter/Deep-Learning-Model-Inversion-Attacks-and-Defenses.

Mobile Edge computing (MEC) in Cyber-Physical Systems (CPS) with massive resource-constrained Edge Computing Node (ECN) faces new challenges in security provisioning. The traditional centralized security authentication schemes with low performance are no longer applied for MEC in CPS.Due to the mobility of ECN, it is extraordinarily practical for ECN to establish a security association with another AP once leaving the service area of its current AP. In this paper, we represent the related research and propose a novel and efficient Software Defined Networking (SDN)-based Handover Authentication Scheme for MEC in CPS (SHAS). An authentication Handover Module (AHM) in the SDN controller is applied for key distribution and authentication management. Before ECN handovers, the AHM distributes a key to the current serving AP for ECN further handover. Whenever a handover happens, target AP requests the AHM for the one-time session key to authenticate the ECN. The target AP and ECN can proceed with the 3-way handshake protocol by the one-time session key to achieve mutual authentication and secret key confidentiality. Using the logical derivation of Burrows, Abadi, and Needham and formal verification by Automated Validation of Internet Security Protocols and Applications (AVISPA), proposed SHAS scheme can get mutual authentication and secret key confidentiality with a strong anti-attack ability. The simulation results show that the SHAS scheme has the characteristics of lower computational delay and less communication resources.Finally, the practical demonstration of our scheme is done using the widely-accepted NS-3 simulation.

With the development of mobile platform, such as smart cellphone and pad, the E-Learning model has been rapidly developed. However, due to the low completion rate for E-Learning platform, it is very necessary to analyze the behavior characteristics of online learners to intelligently adjust online education strategy and enhance the quality of learning. In this paper, we analyzed the relation indicators of E-Learning to build the student profile and gave countermeasures. Adopting the similarity computation and Jaccard coefficient algorithm, we designed a system model to clean and dig into the educational data and also the students’ learning attitude and the duration of learning behavior to establish student profile. According to the E-Learning resources and learner behaviors, we also present the intelligent guide model to guide both E-Learning platform and learners to improve learning things. The study on student profile can help the E-Learning platform to meet and guide the students’ learning behavior deeply and also to provide personalized learning situation and promote the optimization of the E-Learning.

With the rapid development of smart grids, the number of various types of power IoT terminal devices has grown by leaps and bounds. An attack on either of the difficult-to-protect end devices or any node in a large and complex network can put the grid at risk. The traffic generated by Distributed Denial of Service (DDoS) attacks is characterised by short bursts of time, making it difficult to apply existing centralised detection methods that rely on manual setting of attack characteristics to changing attack scenarios. In this paper, a DDoS attack detection model based on Bidirectional Long Short-Term Memory (BiLSTM) is proposed by constructing an edge detection framework, which achieves bi-directional contextual information extraction of the network environment using the BiLSTM network and automatically learns the temporal characteristics of the attack traffic in the original data traffic. This paper takes the DDoS attack in the power Internet of Things as the research object. Simulation results show that the model outperforms traditional advanced models such as Recurrent Neural Network (RNN) and Long Short Term Memory (LSTM) in terms of accuracy, false detection rate, and time delay. It plays an auxiliary role in the security protection of the power Internet of Things and effectively improves the reliability of the power grid.