Lightweight Zero Trust via Automotive SDN
论文针对车载网络信任问题，通过分析现有MACsec/MKA及引入CORECONF/YANG管理方案，提出了一种无需额外基础设施的轻量级零信任架构。
0 citationsRead paper
Selected work
Representative Papers
GraIP: A Benchmarking Framework For Neural Graph Inverse Problems
This work addresses the limitation of existing graph learning methods, which are often designed in isolation for specific tasks and lack a unified framework for tackling the inverse problem of inferring graph structure from observational data. To this end, the authors propose the Neural Graph Inverse Problem (GraIP) framework, which unifies diverse tasks—such as graph structure discovery, causal inference, and neural relational reasoning—as inverse problems of forward processes like message passing or network dynamics. This study presents the first systematic formulation of the GraIP theoretical paradigm, accompanied by a cross-task benchmark dataset and evaluation metrics. Extensive experiments demonstrate the framework’s generality and superior performance over existing baselines across multiple tasks, including graph rewiring, causal discovery, and neural relation inference.
0 citationsRead paper
GraphBench: Next-generation graph learning benchmarking
Graph machine learning has long suffered from benchmark fragmentation: datasets are task-specific, evaluation protocols lack standardization, and out-of-distribution (OOD) generalization is rarely considered—severely hindering reproducibility and cross-model comparison. To address this, we introduce GraphBench, the first cross-domain, multi-task graph learning benchmark platform, supporting node-, edge-, and graph-level classification as well as generative tasks. GraphBench features standardized data splits, a unified evaluation protocol, an automated hyperparameter tuning framework, and—uniquely—integrates OOD generalization metrics into its core evaluation suite. We establish authoritative baselines using message-passing GNNs and graph Transformers, conducting systematic evaluations across 12 diverse datasets. GraphBench significantly improves evaluation consistency and result comparability, providing a reproducible, scalable, and standardized infrastructure for graph learning research.
0 citationsRead paper
From ECU to VSOC: UDS Security Monitoring Strategies
Facing escalating cybersecurity threats targeting the Unified Diagnostic Services (UDS) protocol in modern vehicles, this paper proposes an end-to-end monitoring framework spanning ECU log acquisition, context-aware logging, and collaborative analysis at a remote Vehicle Security Operations Center (VSOC). Methodologically, it introduces a multi-scenario detection architecture grounded in a novel UDS attack taxonomy and designs a lightweight context-correlation analysis technique to significantly improve attack detection accuracy and interpretability. Experimental evaluation demonstrates comprehensive coverage of typical UDS attack vectors—including DoIP abuse and session/security access bypass—with a detection accuracy of 92.3%. Furthermore, the study identifies structural limitations of the AUTOSAR Security Event standard for real-time attack detection and proposes semantic enhancement and standardization extensions for in-vehicle logging. These contributions provide empirical support for the evolution of automotive cybersecurity standards.
0 citationsRead paper
Preference-Based Gradient Estimation for ML-Based Approximate Combinatorial Optimization
Many combinatorial optimization (CO) applications demand rapid generation of high-quality feasible solutions. Method: This paper proposes a data-driven framework that enhances classical approximation algorithms by leveraging graph neural networks (GNNs) to dynamically tune their parameters—while rigorously preserving solution feasibility. Contribution/Results: We introduce preference-based gradient estimation, the first technique enabling end-to-end, self-supervised, differentiable training of black-box approximation algorithms. This bridges the gap between data-adaptive learning and structural guarantees inherent in classical algorithms. Evaluated on the Traveling Salesman Problem (TSP) and Minimum k-Cut, our approach matches state-of-the-art learning-based solvers in solution quality, substantially outperforms the original approximation algorithms, and guarantees feasibility without post-processing.
0 citationsRead paper
Recent publications
Latest Papers
Lightweight Zero Trust via Automotive SDN
论文针对车载网络信任问题，通过分析现有MACsec/MKA及引入CORECONF/YANG管理方案，提出了一种无需额外基础设施的轻量级零信任架构。
0 citationsRead paper
GraIP: A Benchmarking Framework For Neural Graph Inverse Problems
This work addresses the limitation of existing graph learning methods, which are often designed in isolation for specific tasks and lack a unified framework for tackling the inverse problem of inferring graph structure from observational data. To this end, the authors propose the Neural Graph Inverse Problem (GraIP) framework, which unifies diverse tasks—such as graph structure discovery, causal inference, and neural relational reasoning—as inverse problems of forward processes like message passing or network dynamics. This study presents the first systematic formulation of the GraIP theoretical paradigm, accompanied by a cross-task benchmark dataset and evaluation metrics. Extensive experiments demonstrate the framework’s generality and superior performance over existing baselines across multiple tasks, including graph rewiring, causal discovery, and neural relation inference.
0 citationsRead paper
GraphBench: Next-generation graph learning benchmarking
Graph machine learning has long suffered from benchmark fragmentation: datasets are task-specific, evaluation protocols lack standardization, and out-of-distribution (OOD) generalization is rarely considered—severely hindering reproducibility and cross-model comparison. To address this, we introduce GraphBench, the first cross-domain, multi-task graph learning benchmark platform, supporting node-, edge-, and graph-level classification as well as generative tasks. GraphBench features standardized data splits, a unified evaluation protocol, an automated hyperparameter tuning framework, and—uniquely—integrates OOD generalization metrics into its core evaluation suite. We establish authoritative baselines using message-passing GNNs and graph Transformers, conducting systematic evaluations across 12 diverse datasets. GraphBench significantly improves evaluation consistency and result comparability, providing a reproducible, scalable, and standardized infrastructure for graph learning research.
0 citationsRead paper
From ECU to VSOC: UDS Security Monitoring Strategies
Facing escalating cybersecurity threats targeting the Unified Diagnostic Services (UDS) protocol in modern vehicles, this paper proposes an end-to-end monitoring framework spanning ECU log acquisition, context-aware logging, and collaborative analysis at a remote Vehicle Security Operations Center (VSOC). Methodologically, it introduces a multi-scenario detection architecture grounded in a novel UDS attack taxonomy and designs a lightweight context-correlation analysis technique to significantly improve attack detection accuracy and interpretability. Experimental evaluation demonstrates comprehensive coverage of typical UDS attack vectors—including DoIP abuse and session/security access bypass—with a detection accuracy of 92.3%. Furthermore, the study identifies structural limitations of the AUTOSAR Security Event standard for real-time attack detection and proposes semantic enhancement and standardization extensions for in-vehicle logging. These contributions provide empirical support for the evolution of automotive cybersecurity standards.
0 citationsRead paper
Preference-Based Gradient Estimation for ML-Based Approximate Combinatorial Optimization
Many combinatorial optimization (CO) applications demand rapid generation of high-quality feasible solutions. Method: This paper proposes a data-driven framework that enhances classical approximation algorithms by leveraging graph neural networks (GNNs) to dynamically tune their parameters—while rigorously preserving solution feasibility. Contribution/Results: We introduce preference-based gradient estimation, the first technique enabling end-to-end, self-supervised, differentiable training of black-box approximation algorithms. This bridges the gap between data-adaptive learning and structural guarantees inherent in classical algorithms. Evaluated on the Traveling Salesman Problem (TSP) and Minimum k-Cut, our approach matches state-of-the-art learning-based solvers in solution quality, substantially outperforms the original approximation algorithms, and guarantees feasibility without post-processing.
0 citationsRead paper