REFUSE-BENCH
REFUSE‑BENCH is a benchmark dataset created by the University of Maryland, Baltimore for binary function similarity detection. It contains 243,128 binary files compiled under various configurations and optimizations, collected from source code on GitHub to reflect real‑world computer security scenarios. The dataset supports research in reverse engineering, malware analysis, and vulnerability detection, and is used to evaluate and improve binary function similarity models.
Description
Dataset Overview
Dataset Name
- Assemblage
Dataset Description
- Assemblage is a dataset for evaluating function similarity models.
Related Models
- The dataset is used to evaluate the following five models:
- jTrans
- GNN from Li et al.
- Naive Multi‑headed Attention Transformer Encoder
- Ghidra's BSim Plugin
- REFuSe
Dataset Construction
- A recipe and code for building the Assemblage dataset and running BSim experiments are provided.
- Instructions for reproducing the dataset from the recipe can be found here.
Data Processing
- Code for preprocessing data for experiments is included.
Model Training
- Code for training models on the Assemblage data is provided.
Model Evaluation
- Code for evaluating models on the dataset is provided.
AI studio
Generate PPTs instantly with Nano Banana Pro.
Generate PPT NowAccess Dataset
Please login to view download links and access full dataset details.
Topics
Source
Organization: arXiv
Created: 10/30/2024
Power Your Data Analysis with Premium AI Models
Supporting GPT-5, Claude-4, DeepSeek v3, Gemini and more.
Enjoy a free trial and save 20%+ compared to official pricing.