Cross-site scripting (XSS) attacks pose a significant threat to web applications and user privacy, with the number of such attacks rapidly increasing. Although existing machine learning and deep learning-based XSS attack detection models are effective against common XSS attacks, these models all overlook their own security and often fail to defend against adversarial samples that exploit model vulnerabilities, allowing attackers to successfully bypass these models by using XSS adversarial samples. To address this challenge, in this paper, we propose a novel XSS attack detection model based on two-stage Abstract Syntax Tree (AST) analysis and Long Short-Term Memory (LSTM) neural networks, effectively mitigating the impact of adversarial samples. Our model leverages the ability of AST parsing and analysis of HTML and JavaScript code to effectively eliminate redundant information and adversarial perturbations introduced by adversarial samples. The two-stage process first extracts JavaScript code from the HTML AST, then identifies malicious code fragments from the JavaScript AST. Finally, the LSTM neural network is trained to classify samples as malicious or benign. By analyzing the HTML and JavaScript components of web pages, our model identifies and eliminates adversarial perturbations that interfere with detection, significantly enhancing the security and reliability of the detection process. Extensive experiments on real datasets demonstrate our model's superior performance, achieving an accuracy rate of 0.991 and an F1 score of 0.998 against standard XSS samples, outperforming existing models. More importantly, when facing adversarial XSS samples, most existing detection models exhibit severe robustness degradation with the detection rate (DR) below 0.880, whereas our model maintains a detection rate of over 0.982, significantly higher than state-of-the-art models and demonstrating its significant effectiveness in defending against XSS adversarial attacks.

Picture Archiving and Communication System (PACS) is an important technology for the transmission of medical images and related information. Using the PACS technology, medical images can be transmitted and shared over local area networks, wide area networks. However, ensuring transmission security continues to be crucial challenges in the field of modern medicine. In this paper, a hyperbolic memristor model with different hysteresis loops for different input signals is proposed, which has both locally active and passive properties under different parameters. Based on the memristor model, a four-neuron Hopfield neural network (HNN) is introduced, incorporating a memristor as a replacement for one of the synaptic weights, while subjecting one of the four neurons to electromagnetic radiation. The effects of the memristor parameters and coupling weights on neural network are studied, which reveals the existence of coexistence behavior in neural network. In addition, an equivalent circuit of HNN is implemented to demonstrate the accuracy of the dynamical analysis. Finally, a medical image encryption approach with bi-directional dynamic permutation and multi-directional dynamic diffusion process is proposed. The experimental findings demonstrate that the encryption scheme has strong robustness, which can be applied in PACS to enhance the security of medical images during transmission.

With the continuous development of computers, communication technology, and regional medical collaboration services, the security and confidentiality of information are becoming more and more important. In order to prevent the illegal leakage of sensitive patient information, it is of great significance to study medical image encryption. In this article, a flux-controlled hyperbolic memristor model with locally active characteristics is proposed, which has rich nonlinear characteristics. The memristor parameters affect the local activity of the memristor, which is explained by mathematical analysis. Based on the traditional hopfield neural network (HNN), a memristive fully connect neural network (MFNN) containing four neurons is constructed with more complex coupling relationships between individual neurons. The memristor can be used to characterize the effect of external electromagnetic radiation on neurons. The complex dynamical behaviors of MFNN are found by numerical simulations. An equivalent circuit for the neural network is constructed to verify the accuracy of the numerical simulation. In addition, a medical image encryption scheme based on MFNN is proposed. The encryption scheme performs a bit-level permutation of the original image using a chaotic sequence randomly generated by the chaotic system. Fibonacci$Q$-matrix and central diffusion algorithm are used to diffuse the permutation image. Through numerical analysis, the maximum entropy of this encryption algorithm reaches 7.99, and the correlation is close to zero, which proves the resistance of the algorithm to statistical attacks. The algorithm takes only 3.9 s to encrypt an 8-bit medical image of 320 × 320 size on Windows 10 operating system. Experimental results show that the proposed encryption scheme is very secure and has good applications in medical image encryption.