Selected work

Representative Papers

Tourists Profiling by Interest Analysis

Dec 05, 2025International Conference on Advanced Data Mining and Applications

Traditional statistical methods struggle to uncover the underlying motivations behind tourist behavior and interest evolution within attraction networks. Method: We propose the first integrated analytical framework that jointly models semantic features (via LDA topic modeling of textual digital footprints) and structural features (via visit-sequence graph modeling), enhanced by graph neural networks (GNNs) and multi-source trajectory clustering. This enables interpretable tourist interest profiling and cross-city mobility mapping. Contribution/Results: Evaluated on real-world tourism datasets, our approach achieves a 23.6% improvement in interest cluster identification accuracy. It significantly advances understanding of how tourist preferences form and shift over time and space. By bridging semantic intent with spatiotemporal behavioral structure, the framework establishes a novel paradigm for intelligent tourism recommendation and collaborative destination management.

2 citationsRead paper

Analytic Regularity and Approximation Limits of Coefficient-Constrained Shallow Networks

Jan 08, 2026

This work investigates the approximation capabilities of single-hidden-layer neural networks with globally constrained coefficients—such as those bounded in ℓ¹ norm or exhibiting subexponential growth—when targeting non-analytic functions. Through a deterministic analysis that combines comparison arguments with Bernstein-type estimates, the study demonstrates that such networks remain “rigid” under Gevrey-class activation functions: their approximation error is fundamentally governed by the best polynomial approximation rate, with only exponentially small residual terms. The findings establish that, even when employing non-analytic activations, shallow networks with coefficient constraints cannot surpass classical polynomial approximation rates, thereby revealing an intrinsic limitation in their representational power.

0 citationsRead paper

Systemic approach for modeling a generic smart grid

Nov 21, 2025

Interdisciplinary modeling and large-scale simulation of smart grids face challenges including model heterogeneity, complex inter-domain couplings, and poor computational scalability. To address these, this paper proposes a systematic, multi-domain co-modeling framework that integrates power system dynamics, energy market mechanisms, and demand-side response behaviors into a unified backbone model. It introduces an innovative distributed subsystem optimization architecture to support flexible and scalable prosumer-coordinated scheduling. By unifying system-level modeling, distributed optimization, and cross-domain simulation techniques, the framework enables integrated modeling and co-simulation of heterogeneous, multi-source resources. The developed simulation tool efficiently validates diverse grid evolution scenarios, enabling system-level hypothesis testing at human timescales. Empirical evaluation demonstrates a 37% improvement in modeling efficiency and a 22% reduction in error for critical scenarios.

0 citationsRead paper

OpenGL GPU-Based Rowhammer Attack (Work in Progress)

Sep 24, 2025

Rowhammer attacks exploit high-frequency memory access to induce bit flips in adjacent DRAM rows, compromising memory integrity. This paper proposes an adaptive, multi-faceted Rowhammer attack leveraging GPU compute shaders: it employs OpenGL compute shaders to achieve fine-grained, highly concurrent memory access; introduces a statistics-driven target-row selection strategy and dynamic parameter adjustment mechanism to circumvent existing hardware and software mitigations; and integrates memory pattern initialization, iterative row hammering, and real-time error detection for low-overhead, efficient bit-flip triggering. Evaluated on the Raspberry Pi 4 platform, the GPU-accelerated approach achieves significantly higher bit-flip rates than CPU-based methods. This work constitutes the first systematic demonstration of GPU-accelerated Rowhammer, establishing its feasibility and heightened destructive capability. It provides a novel paradigm for hardware security evaluation and highlights critical vulnerabilities in GPU-mediated memory access patterns.

0 citationsRead paper

Recent publications

Latest Papers

Analytic Regularity and Approximation Limits of Coefficient-Constrained Shallow Networks

Jan 08, 2026

This work investigates the approximation capabilities of single-hidden-layer neural networks with globally constrained coefficients—such as those bounded in ℓ¹ norm or exhibiting subexponential growth—when targeting non-analytic functions. Through a deterministic analysis that combines comparison arguments with Bernstein-type estimates, the study demonstrates that such networks remain “rigid” under Gevrey-class activation functions: their approximation error is fundamentally governed by the best polynomial approximation rate, with only exponentially small residual terms. The findings establish that, even when employing non-analytic activations, shallow networks with coefficient constraints cannot surpass classical polynomial approximation rates, thereby revealing an intrinsic limitation in their representational power.

0 citationsRead paper

Tourists Profiling by Interest Analysis

Dec 05, 2025International Conference on Advanced Data Mining and Applications

Traditional statistical methods struggle to uncover the underlying motivations behind tourist behavior and interest evolution within attraction networks. Method: We propose the first integrated analytical framework that jointly models semantic features (via LDA topic modeling of textual digital footprints) and structural features (via visit-sequence graph modeling), enhanced by graph neural networks (GNNs) and multi-source trajectory clustering. This enables interpretable tourist interest profiling and cross-city mobility mapping. Contribution/Results: Evaluated on real-world tourism datasets, our approach achieves a 23.6% improvement in interest cluster identification accuracy. It significantly advances understanding of how tourist preferences form and shift over time and space. By bridging semantic intent with spatiotemporal behavioral structure, the framework establishes a novel paradigm for intelligent tourism recommendation and collaborative destination management.

2 citationsRead paper

Systemic approach for modeling a generic smart grid

Nov 21, 2025

Interdisciplinary modeling and large-scale simulation of smart grids face challenges including model heterogeneity, complex inter-domain couplings, and poor computational scalability. To address these, this paper proposes a systematic, multi-domain co-modeling framework that integrates power system dynamics, energy market mechanisms, and demand-side response behaviors into a unified backbone model. It introduces an innovative distributed subsystem optimization architecture to support flexible and scalable prosumer-coordinated scheduling. By unifying system-level modeling, distributed optimization, and cross-domain simulation techniques, the framework enables integrated modeling and co-simulation of heterogeneous, multi-source resources. The developed simulation tool efficiently validates diverse grid evolution scenarios, enabling system-level hypothesis testing at human timescales. Empirical evaluation demonstrates a 37% improvement in modeling efficiency and a 22% reduction in error for critical scenarios.

0 citationsRead paper

OpenGL GPU-Based Rowhammer Attack (Work in Progress)

Sep 24, 2025

Rowhammer attacks exploit high-frequency memory access to induce bit flips in adjacent DRAM rows, compromising memory integrity. This paper proposes an adaptive, multi-faceted Rowhammer attack leveraging GPU compute shaders: it employs OpenGL compute shaders to achieve fine-grained, highly concurrent memory access; introduces a statistics-driven target-row selection strategy and dynamic parameter adjustment mechanism to circumvent existing hardware and software mitigations; and integrates memory pattern initialization, iterative row hammering, and real-time error detection for low-overhead, efficient bit-flip triggering. Evaluated on the Raspberry Pi 4 platform, the GPU-accelerated approach achieves significantly higher bit-flip rates than CPU-based methods. This work constitutes the first systematic demonstration of GPU-accelerated Rowhammer, establishing its feasibility and heightened destructive capability. It provides a novel paradigm for hardware security evaluation and highlights critical vulnerabilities in GPU-mediated memory access patterns.

0 citationsRead paper