Selected work

Representative Papers

HTG-GCL: Leveraging Hierarchical Topological Granularity from Cellular Complexes for Graph Contrastive Learning

Nov 30, 2025

Existing graph contrastive learning (GCL) methods struggle to identify task-relevant topological structures and fail to adaptively learn multi-granular topological representations required by downstream tasks. To address this, we propose HTG-GCL—a hierarchical topological graph GCL framework. First, we introduce the novel concept of *topological granularity* and construct multi-scale contrastive graph views grounded in cycle-based cell complexes. Second, we design a multi-granularity decoupled contrastive mechanism that jointly models coarse-grained global structure and fine-grained local patterns. Third, we propose an uncertainty-aware granularity weighting strategy to dynamically fuse hierarchical topological information. Extensive experiments on multiple benchmark datasets demonstrate that HTG-GCL consistently outperforms state-of-the-art GCL methods, validating its effectiveness in enhancing representation discriminability, task adaptability, and robustness.

0 citationsRead paper

Interleaved Learning and Exploration: A Self-Adaptive Fuzz Testing Framework for MLIR

Oct 09, 2025

Existing MLIR compiler correctness testing suffers from insufficient test case diversity and low semantic validity. Method: This paper proposes the first adaptive fuzzing framework that integrates neural program generation with dynamic feedback-driven learning. It introduces perturbation-based sampling to enhance input diversity, employs syntax–semantics co-learning and a dual-path model optimization mechanism to autonomously generate high-quality, semantically valid test programs from minimal seeds, and iteratively refines the model via crash/non-crash feedback loops—overcoming inherent limitations of template- or rule-based approaches. Contribution/Results: Within 30 days, the framework discovers 80 previously unknown vulnerabilities; within 24 hours, it detects 53 bugs—more than 3.5× the best baseline. It achieves 28.2% code coverage, improving upon the state-of-the-art tool by 42%.

0 citationsRead paper

Exploring the Feasibility of End-to-End Large Language Model as a Compiler

Jun 30, 2025IEEE International Joint Conference on Neural Network

This work investigates the feasibility of the “Large Language Model as a Compiler” (LaaC) paradigm—i.e., whether LLMs can perform end-to-end, precise compilation from source code to target assembly. To this end, the authors introduce CompilerEval†, the first benchmark dataset specifically designed for compilation tasks, comprising multilingual source code paired with cross-platform (x86, ARM, RISC-V) assembly. They develop a dedicated evaluation framework and employ prompt engineering, chain-of-thought reasoning, and model scaling to systematically assess the source-code understanding and assembly-generation capabilities of leading open- and closed-source LLMs. Results show that current LLMs possess foundational compilation ability; targeted optimizations substantially improve assembly correctness and overall compilation success rates. This study provides the first systematic empirical validation of LaaC’s technical viability, proposes principled architectural guidelines and evolutionary pathways for compilation-oriented LLMs, and establishes a foundation for AI-native compiler research.

0 citationsRead paper

HACMony: Automatically Testing Hopping-related Audio-stream Conflict Issues on HarmonyOS

Apr 10, 2025

This paper addresses the Audio-stream Conflict (HAC) problem in HarmonyOS distributed scenarios, wherein ambiguous cross-device application navigation semantics and the absence of multi-device collaborative testing mechanisms hinder effective HAC detection. To tackle this, we propose the first formal audio navigation semantics model and construct an Audio Service Transition Graph (ASTG). Based on ASTG, we design a model-driven automated HAC detection method. We systematically identify and categorize two canonical HAC patterns: MOD (Multi-device Concurrent Preemption) and MOR (Migration State Mismatch). Our approach detects HACs in 11 out of 20 real-world HarmonyOS applications, enabling precise localization and classification. The framework provides a reusable diagnostic infrastructure for both system and application developers, along with empirically grounded insights for mitigating distributed audio inconsistencies.

0 citationsRead paper

Recent publications

Latest Papers

HTG-GCL: Leveraging Hierarchical Topological Granularity from Cellular Complexes for Graph Contrastive Learning

Nov 30, 2025

Existing graph contrastive learning (GCL) methods struggle to identify task-relevant topological structures and fail to adaptively learn multi-granular topological representations required by downstream tasks. To address this, we propose HTG-GCL—a hierarchical topological graph GCL framework. First, we introduce the novel concept of *topological granularity* and construct multi-scale contrastive graph views grounded in cycle-based cell complexes. Second, we design a multi-granularity decoupled contrastive mechanism that jointly models coarse-grained global structure and fine-grained local patterns. Third, we propose an uncertainty-aware granularity weighting strategy to dynamically fuse hierarchical topological information. Extensive experiments on multiple benchmark datasets demonstrate that HTG-GCL consistently outperforms state-of-the-art GCL methods, validating its effectiveness in enhancing representation discriminability, task adaptability, and robustness.

0 citationsRead paper

Interleaved Learning and Exploration: A Self-Adaptive Fuzz Testing Framework for MLIR

Oct 09, 2025

Existing MLIR compiler correctness testing suffers from insufficient test case diversity and low semantic validity. Method: This paper proposes the first adaptive fuzzing framework that integrates neural program generation with dynamic feedback-driven learning. It introduces perturbation-based sampling to enhance input diversity, employs syntax–semantics co-learning and a dual-path model optimization mechanism to autonomously generate high-quality, semantically valid test programs from minimal seeds, and iteratively refines the model via crash/non-crash feedback loops—overcoming inherent limitations of template- or rule-based approaches. Contribution/Results: Within 30 days, the framework discovers 80 previously unknown vulnerabilities; within 24 hours, it detects 53 bugs—more than 3.5× the best baseline. It achieves 28.2% code coverage, improving upon the state-of-the-art tool by 42%.

0 citationsRead paper

Exploring the Feasibility of End-to-End Large Language Model as a Compiler

Jun 30, 2025IEEE International Joint Conference on Neural Network

This work investigates the feasibility of the “Large Language Model as a Compiler” (LaaC) paradigm—i.e., whether LLMs can perform end-to-end, precise compilation from source code to target assembly. To this end, the authors introduce CompilerEval†, the first benchmark dataset specifically designed for compilation tasks, comprising multilingual source code paired with cross-platform (x86, ARM, RISC-V) assembly. They develop a dedicated evaluation framework and employ prompt engineering, chain-of-thought reasoning, and model scaling to systematically assess the source-code understanding and assembly-generation capabilities of leading open- and closed-source LLMs. Results show that current LLMs possess foundational compilation ability; targeted optimizations substantially improve assembly correctness and overall compilation success rates. This study provides the first systematic empirical validation of LaaC’s technical viability, proposes principled architectural guidelines and evolutionary pathways for compilation-oriented LLMs, and establishes a foundation for AI-native compiler research.

0 citationsRead paper

HACMony: Automatically Testing Hopping-related Audio-stream Conflict Issues on HarmonyOS

Apr 10, 2025

This paper addresses the Audio-stream Conflict (HAC) problem in HarmonyOS distributed scenarios, wherein ambiguous cross-device application navigation semantics and the absence of multi-device collaborative testing mechanisms hinder effective HAC detection. To tackle this, we propose the first formal audio navigation semantics model and construct an Audio Service Transition Graph (ASTG). Based on ASTG, we design a model-driven automated HAC detection method. We systematically identify and categorize two canonical HAC patterns: MOD (Multi-device Concurrent Preemption) and MOR (Migration State Mismatch). Our approach detects HACs in 11 out of 20 real-world HarmonyOS applications, enabling precise localization and classification. The framework provides a reusable diagnostic infrastructure for both system and application developers, along with empirically grounded insights for mitigating distributed audio inconsistencies.

0 citationsRead paper