EMBER2024
收藏资源简介:
EMBER2024是一个新的数据集,旨在全面评估恶意软件分类器。该数据集由EMBER2017和EMBER2018的作者与相关机构合作创建,包含来自六种文件格式的超过320万文件的哈希、元数据、特征向量和标签。数据集支持对七个恶意软件分类任务的机器学习模型的训练和评估,包括恶意软件检测、恶意软件家族分类和恶意软件行为识别。EMBER2024是第一个包含最初未被任何防病毒产品检测到的恶意文件集合的数据集,创建了一个“挑战”集,以评估分类器对逃避恶意软件的性能。该数据集还引入了EMBER特征版本3,增加了对新特征类型的支持。
EMBER2024 is a novel dataset designed to comprehensively evaluate malware classifiers. Developed in collaboration with the authors of EMBER2017 and EMBER2018 and relevant institutions, this dataset contains hashes, metadata, feature vectors, and labels for over 3.2 million files across six file formats. The dataset supports the training and evaluation of machine learning models for seven malware classification tasks, including malware detection, malware family classification, and malware behavior recognition. EMBER2024 is the first dataset to include a collection of malicious files that were initially undetected by any antivirus products, creating a "challenge set" to assess the performance of classifiers against evasive malware. The dataset also introduces EMBER Feature Version 3, which adds support for new feature types.

- 1通过Booz Allen Hamilton, Laboratory for Physical Sciences, CrowdStrike, Cisco Systems · 2025年



