SearcharxivSearch

arXiv subjects

Le Cong Trinh

Publications and source records attributed to Le Cong Trinh.

2 recordsLinked to original sources

On directional second-order tangent sets of analytic sets and applications in optimization

In this paper we study directional second-order tangent sets of real and complex analytic sets. For an analytic set $X\subseteq \mathbb K^n$ and a nonzero tangent direction $u\in T_0X$, we compare the geometric directional second-order tangent set $T^2_{0,u}X$, defined through second-order expansions of analytic curves in $X$, with the algebraic directional second-order tangent set $T^{2,a}_{0,u}X$, defined by the initial forms of the equations of $X$. We first prove the general inclusion $T^2_{0,u}X\subseteq T^{2,a}_{0,u}X$ and exhibit explicit real and complex analytic examples showing that this inclusion can be strict. These examples show that algebraically admissible second-order coefficients need not be geometrically realizable by analytic curves in $X$. To address this gap, we reformulate the equality $T^2_{0,u}X=T^{2,a}_{0,u}X$ as a realizability problem: the two sets coincide whenever every algebraically admissible second-order coefficient is realized by an analytic curve in $X$ with prescribed first two terms. We establish this realizability property for several important classes of analytic sets, including smooth analytic germs, homogeneous analytic cones, hypersurfaces with nondegenerate tangent directions, and nondegenerate analytic complete intersections. As an application, we derive second-order necessary and sufficient optimality conditions for $C^2$ optimization problems on closed sets. In the analytic setting, whenever the above equality holds, the geometric directional second-order tangent sets appearing in these conditions may be replaced by their algebraic counterparts, so that the second-order tests become explicitly computable from the defining equations of the feasible set.

math.AG

VulnSense: Efficient Vulnerability Detection in Ethereum Smart Contracts by Multimodal Learning with Graph Neural Network and Language Model

This paper presents VulnSense framework, a comprehensive approach to efficiently detect vulnerabilities in Ethereum smart contracts using a multimodal learning approach on graph-based and natural language processing (NLP) models. Our proposed framework combines three types of features from smart contracts comprising source code, opcode sequences, and control flow graph (CFG) extracted from bytecode. We employ Bidirectional Encoder Representations from Transformers (BERT), Bidirectional Long Short-Term Memory (BiLSTM) and Graph Neural Network (GNN) models to extract and analyze these features. The final layer of our multimodal approach consists of a fully connected layer used to predict vulnerabilities in Ethereum smart contracts. Addressing limitations of existing vulnerability detection methods relying on single-feature or single-model deep learning techniques, our method surpasses accuracy and effectiveness constraints. We assess VulnSense using a collection of 1.769 smart contracts derived from the combination of three datasets: Curated, SolidiFI-Benchmark, and Smartbugs Wild. We then make a comparison with various unimodal and multimodal learning techniques contributed by GNN, BiLSTM and BERT architectures. The experimental outcomes demonstrate the superior performance of our proposed approach, achieving an average accuracy of 77.96\% across all three categories of vulnerable smart contracts.

cs.CR