Huiyan Zhang

The accelerated development of artificial intelligence, big data, and the Internet of Things has profoundly expanded the scale, sources, and application scenarios of data, which increasingly exhibit complex characteristics such as high dimensionality, heterogeneity, dynamics, and multimodality

This article proposes a novel observer-based adaptive neural network-based resilient consensus control approach to address hybrid cyberattacks, disturbances, and nonlinear dynamics in nonlinear leader-following multiagent systems (MASs). Specifically, a dimension expansion methodology is developed to dynamically model and compensate for false data injection (FDI) attacks, while denial-of-service (DoS) attacks are probabilistically characterized via Bernoulli variables, forming a comprehensive hybrid attack mitigation strategy. Then, a cascaded observer is designed, integrating dimension-extended system modeling with disturbance decoupling to simultaneously estimate system states and external disturbances with high precision. Furthermore, an adaptive neural network-based approximation scheme is employed to handle system nonlinearities, eliminating the conservatism of Lipschitz-based methods while enhancing robustness in complex environments. Finally, the simulation result validates that the proposed control method achieves resilient consensus of leader-following MASs under hybrid cyberattacks, disturbances, and nonlinear dynamics.

Deep neural networks have recently achieved promising performance in the vein recognition task and have shown an increasing application trend. However, they are prone to adversarial attacks by adding imperceptible perturbations to the input, resulting in incorrect recognition. To address this issue, we propose a novel defense model named MsMemoryGAN, which aims to filter the perturbations from adversarial samples before recognition. First, we design a multiscale memory autoencoder (MsMemoryAE) to achieve high-quality reconstruction, where the memory module (MM) within it is capable of learning the detailed patterns of normal samples at different scales. Second, to overcome the limitations of handcrafted similarity metrics, we propose an MM with learnable similarity (LSMM), which retrieves the most relevant memory items to purify the input feature. Finally, the perceptual loss and adversarial loss are integrated with the pixel loss to further enhance the quality of the reconstructed image. During the training phase, the MsMemoryGAN learns to reconstruct the input by merely using fewer prototypical elements of the normal patterns recorded in the memory. At the testing stage, given an adversarial sample, the MsMemoryGAN retrieves its most relevant normal patterns in MMs for reconstruction. Perturbations in the adversarial sample are usually not reconstructed well, resulting in adversarial purification. We conduct extensive experiments on two public vein datasets under different adversarial attack methods to evaluate the performance of the proposed approach. The experimental results show that our approach removes a wide variety of adversarial perturbations, allowing vein classifiers to achieve the highest recognition accuracy.

In this paper, secure control strategies are developed for stochastically sampled systems under Denial-of-Service (DoS)/deception attacks. The actual sampling intervals deviate from their nominal values, and these sampling inaccuracies are modeled as uniformly bounded random variables following a truncated Gaussian distribution. Unlike existing studies on sampled-data systems affected by attack-induced packet losses, the obtained stability condition can be directly utilized regardless of variations in the maximum number of consecutive DoS attacks. We begin by developing a continuous-time modeling framework that enables stabilization analysis for the stochastically sampled systems exposed to DoS and deception attacks. To ensure stochastic stability and facilitate secure control, a discrete-time reformulation of the continuous-time model is then introduced. It is shown that the continuous-time system is stochastically stable if and only if its discrete-time counterpart maintains stochastic stability. Following this, a stability condition is derived and a controller synthesis algorithm is proposed to guarantee the stochastic stability of the continuous-time stochastically sampled systems subject to DoS/deception attacks. Finally, simulation results are presented to demonstrate the efficacy of the proposed controller algorithm.

In this paper, stabilizing control function is designed for nonlinear strict-feedback stochastic systems driven by time delays and Lévy noise. Backstepping method is employed to design the control function which globally asymptotically stabilizes the system at the origin in probability. Further, sufficient conditions are derived for the fourth-moment exponential stability of the system using Itô’s differential of the Lyapunov functional. A numerical example and application of the theoretical results to a practical system are presented. Further, the effectiveness of the designed control function is shown by depicting the backstepping error covariance matrix representation in the simulation results.