API calls for malware detection
收藏资源简介:
本研究创建了迄今为止最大的公开可用的API调用数据集,包含超过30万个恶意软件样本和1万个良性软件样本的API调用实例。数据集基于当前恶意软件和良性软件样本,未压缩大小超过550GB,可在Zenodo上获取。该数据集旨在解决当前机器学习模型在恶意软件检测中需要大量数据的需求,同时提供了一种轻量级的恶意软件检测模型,该模型基于API调用,无需考虑调用顺序,具有较高的准确性和可扩展性。
This study constructs the largest publicly available API call dataset to date, which includes API call instances from over 300,000 malicious software samples and 10,000 benign software samples. Based on current malicious and benign software specimens, the uncompressed size of this dataset exceeds 550 GB, and it is accessible on Zenodo. This dataset is designed to address the substantial data requirements of contemporary machine learning models for malware detection. Additionally, it provides a lightweight malware detection model that relies on API calls without the need to consider call order, boasting high accuracy and strong scalability.

- 1Malware Detection based on API calls德国帕绍大学, 冰岛雷克雅未克大学 · 2025年



