In recent years, the number of malicious web pages has increased dramatically, posing a great challenge to network security. While current machine learning-based detection methods have emerged as a promising alternative to traditional detection techniques. However, these methods are commonly based on single-modal features or simple stacking of classifiers built on various features. As a result, these techniques are not capable of effectively fusing features from different modalities, ultimately limiting the detection effectiveness. To address this limitation, we propose a malicious web page detection method based on multi-modal learning and pre-trained models. First, in the input stage, the raw URL and HTML tag sequences of web pages are used as input features. To help the subsequent model learn the relationship between the two modalities and avoid information confusion, modal-type encoding, and positional encoding are introduced. Next, a single-stream neural network based on the ConvBERT pre-trained model is used as the backbone classifier, and it learns the representation of multi-modal features through fine-tuning. For the output part of the model, a linear layer based on large margin softmax is applied to the decision-making. This activation function effectively increases the classification boundary and improves the robustness. In addition, a coarse-grained modal matching loss is added to the model optimization objective to assist the models in learning the cross-modal association features. Experimental results on synthetic datasets show that our proposed method outperforms traditional single-modal detection methods in general, and has advantages over baseline models in terms of accuracy and reliability.

Cyber security situation awareness,has become a hotpot of research. However, the existing cyber security situation awareness methods are difficult to extract high-order features from network traffic data. In this work, we present an improved cyber security situation awareness method based on ResMLP and LSTM network from a new perspective. Our work focus on cyber attack behavior analysis, that is a key research content of cyber security situation awareness. It introduces the Residual Multi-Layer Perceptrons in deep learning into the network structure of long-short term memory. It can effectively extract the spatial and temporal characteristics of network traffic data, reduce the computational complexity, and improve the accuracy of cyber security situation awareness. Firstly, we extract the spatial features using the ResMLP network. Secondly, we extract the temporal characteristics using the LSTM network. The architecture of the ResMLP network replaces the self-noticing layer with a linear interaction layer, and this design architecture allows the model to guarantee accurate cyber attack behavior analysis performance while balancing the computational cost of the model, which can improve the detection efficiency of the model. Considering that the network data are fed into the model in the form of time series after processing, the model incorporates LSTM networks to avoid the gradient problem while better bringing up the temporal characteristics in the data.The experimental results show that the proposed method can model the future cyber security situation in a network environment more accurately than other similar methods.

Modern physical protection systems (PPSs) play a pivotal role in safeguarding critical infrastructure and maintaining public safety. Yet increasingly complex system architectures and evolving threat landscapes pose significant vulnerability challenges to PPSs. Conventional vulnerability assessment methods predominantly rely on expert knowledge or single-path analysis, which inadequately captures complex inter-component relationships and the impact of uncertainties on PPS vulnerabilities. To bridge this gap, this paper introduces a hybrid analytical framework synergizing complex network theory with fuzzy Petri net (FPN). The proposed method operates through two integrated phases: (1) constructing topological models of PPS using complex network theory to characterize component interrelationships, and (2) incorporating FPN to establish vulnerability propagation models that simulate cascading effects and quantify overall system vulnerability. Compared with conventional methods, the proposed approach demonstrates superior effectiveness in identifying critical vulnerability points within the system, providing a scientifically grounded foundation for enhancing PPS security and implementing risk control measures.

Graph anomaly detection technology has broad applications in financial fraud and risk control. However, existing graph anomaly detection methods often face significant challenges when dealing with complex and variable abnormal patterns, as anomalous nodes are often disguised and mixed with normal nodes, leading to the coexistence of homophily and heterophily in the graph domain. Recent spectral graph neural networks have made notable progress in addressing this issue; however, current techniques typically employ fixed, globally shared filters. This 'one-size-fits-all' approach can easily cause over-smoothing, erasing critical high-frequency signals needed for fraud detection, and lacks adaptive capabilities for different graph instances. To solve this problem, we propose a Multi-Head Spectral-Adaptive Graph Neural Network (MHSA-GNN). The core innovation is the design of a lightweight hypernetwork that, conditioned on a 'spectral fingerprint' containing structural statistics and Rayleigh quotient features, dynamically generates Chebyshev filter parameters tailored to each instance. This enables a customized filtering strategy for each node and its local subgraph. Additionally, to prevent mode collapse in the multi-head mechanism, we introduce a novel dual regularization strategy that combines teacher-student contrastive learning (TSC) to ensure representation accuracy and Barlow Twins diversity loss (BTD) to enforce orthogonality among heads. Extensive experiments on four real-world datasets demonstrate that our method effectively preserves high-frequency abnormal signals and significantly outperforms existing state-of-the-art methods, especially showing excellent robustness on highly heterogeneous datasets.

Large Language Models (LLMs) have demonstrated significant potential and effectiveness across multiple application domains. To assess the performance of mainstream LLMs in public security tasks, this study aims to construct a specialized evaluation benchmark tailored to the Chinese public security domain--CPSDbench. CPSDbench integrates datasets related to public security collected from real-world scenarios, supporting a comprehensive assessment of LLMs across four key dimensions: text classification, information extraction, question answering, and text generation. Furthermore, this study introduces a set of innovative evaluation metrics designed to more precisely quantify the efficacy of LLMs in executing tasks related to public security. Through the in-depth analysis and evaluation conducted in this research, we not only enhance our understanding of the performance strengths and limitations of existing models in addressing public security issues but also provide references for the future development of more accurate and customized LLM models targeted at applications in this field.