Research map / Large language models / Language, safety and interpretability
LC3: Long Cross-Language Code Clone Detection Enhanced by Opcode Sequences and Affinity Aggregation
Xilin Lan, Huan Zhang, Yang Yang, Chengwu Xue, Li Kuang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add opcode sequence modality to source-code encoder (ablation) | LC3-Dual · CSNCC | MAP: 60.38 -> 65.96 | Improved |
| add cross-attention fusion (ablation) | LC3-Dual · CSNCC | MAP: 58.27 -> 65.96 | Improved |
| add adaptive gating (ablation) | LC3-Dual · CSNCC | MAP: 62.65 -> 65.96 | Improved |
| add cycle consistency loss (ablation) | LC3-Dual · CSNCC | MAP: 63.72 -> 65.96 | Improved |
| add adversarial alignment loss (ablation) | LC3-Dual · CSNCC | MAP: 65.57 -> 65.96 | Improved |
| use opcode sequences alone without source code (ablation) | LC3-Dual · CSNCC | MAP: 65.96 -> 2.82 | Hurt |
| replace truncation with SAA on >1024-token code (ablation) | LC3-Dual · CodeNet-VL (>1024 tokens) | MAP: 49.29 -> 63.13 | Improved |
| replace truncation with SAA on >1024-token code | ZC3 · CodeNet-VL (>1024 tokens) | MAP: 46.15 -> 61.53 | Improved |
| replace truncation with SAA on 512-1024 token code (ablation) | LC3-Dual · CodeNet-VL (512,1024] tokens | MAP: 73.13 -> 76.83 | Improved |
| use representation-level pooling instead of truncation (ablation) | LC3-Dual · CodeNet-VL (>1024 tokens) | MAP: 49.29 -> 46.20 (mean), 43.89 (max) | Hurt |
| average block-pair scores instead of max affinity (ablation) | LC3-Dual · CodeNet-VL (>1024 tokens) | MAP: 62.22 (MAS) -> 45.54 | Hurt |
| add contextual support score to MAS (ablation) | LC3-Dual SAA · CodeNet-VL (>1024 tokens) | MAP: 62.22 -> 63.13 | Improved |
Closest papers by meaning
- AdaCCD: Adaptive Semantic Contrasts Discovery Based Cross Lingual Adaptation for Code Clone Detection (AAAI 2024 · similarity 0.89)
- Uncovering LLM-Generated Code: A Zero-Shot Synthetic Code Detector via Code Rewriting (AAAI 2025 · similarity 0.86)
- LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Code (ICLR 2025 · similarity 0.84)
- CODE REPRESENTATION LEARNING AT SCALE (ICLR 2024 · similarity 0.84)
- CoRNStack: High-Quality Contrastive Data for Better Code Retrieval and Reranking (ICLR 2025 · similarity 0.84)
- AST-T5: Structure-Aware Pretraining for Code Generation and Understanding (ICML 2024 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
