模拟真实网络流量的入侵检测数据集

CIC-IDS2017 数据集是一个模拟真实网络流量的入侵检测数据集，涵盖了多种常见协议（如 HTTP、HTTPS、FTP、SSH）以及五种典型的网络攻击类型：DoS（拒绝服务攻击）、DDoS（分布式拒绝服务攻击）、Web 攻击（如 SQL 注入、XSS）、恶意软件攻击（如 Botnets）和信息收集攻击（如端口扫描）。数据集来源于公开资源，包含五个 CSV 文件，分别对应不同攻击类型的网络流量数据。该数据集完全共享，适用于测试网络入侵检测模型的性能，尤其在隐私保护异常检测领域具有重要价值。在论文《Privacy-preserving Industrial Control System Anomaly Detection Platform》中，该数据集被用于验证隐私保护决策树推理协议的时间开销和准确率。数据集于 2023 年 6 月在浙江大学网络空间安全实验室进行了测试使用。

The CIC-IDS2017 dataset is an intrusion detection dataset that simulates real-world network traffic, covering a variety of common protocols such as HTTP, HTTPS, FTP, SSH, and five typical types of network attacks: DoS (Denial of Service), DDoS (Distributed Denial of Service), Web attacks including SQL injection and XSS, malicious software attacks such as Botnets, and information gathering attacks like port scanning. The dataset is sourced from public resources and contains five CSV files corresponding to network traffic data of different attack types respectively. This dataset is fully shared, suitable for testing the performance of network intrusion detection models, and holds particular value in the field of privacy-preserving anomaly detection. In the paper titled "Privacy-preserving Industrial Control System Anomaly Detection Platform", this dataset was used to verify the time overhead and accuracy of the privacy-preserving decision tree inference protocol. The dataset was tested and utilized in the Cyberspace Security Laboratory of Zhejiang University in June 2023.

搜集汇总

背景与挑战

背景概述

该数据集是CIC-IDS2017入侵检测数据集，模拟真实网络流量，涵盖多种常见协议和五种典型网络攻击类型，包括DoS、DDoS、Web攻击等。数据集包含五个CSV文件，数据量466.88MB，完全共享，适用于测试网络入侵检测模型性能，尤其在隐私保护异常检测领域具有重要价值，源自浙江大学的国家重点研发计划项目。