dataset of intrusion detection with SYSLOG FILE for train neural network. [closed]

Question

can I have a link of website where I can download a dataset of attack on network for train my Neural Network whom purpose is to detect an eventual attack on network ? Thx for your help.

Answer 1

There is actually a new intrusion detection dataset with labeled data and full packet capture called the ISCX 2012 dataset available for download via: http://www.iscx.ca/dataset. You just need to fill in the form and submit a request. Since the data is labeled, you can easily use it for training and testing your neural network or any other type of algorithm. The dataset contains over 80GByte of data in pcap format captured over a 7 day span with multiple attacks, as well as normal traffic. Check the details at http://www.iscx.ca/dataset

The DARPA dataset and its derivate, the KDD 99 dataset, are very outdated.

There are 3 days of traffic with normal network activity than can be used for training purposes and 4 days of network activity that includes complex multi-step attacks, each performed on a separate day. These 4 days can be used for testing purposes. Each day of network activity is captured on a separate pcap file for easier analysis.

Friday: Normal Activity. No malicious activity
Saturday: Normal Activity. No malicious activity
Sunday: Infiltrating the network from inside + Normal Activity
Monday: HTTP Denial of Service + Normal Activity
Tuesday: Distributed Denial of Service using an IRC Botnet
Wednesday: Normal Activity. No malicious activity
Thursday: Brute Force SSH + Normal Activity

Answer 2

There is no such public data set exists. Only one data set comes close that is KDD99, DARPA dataset. If you find such a data set , please do inform us. I would like to get it.

See also about IEEE VAST Challenge 2011

Challange

See about KDD99 http://www.kdnuggets.com/news/2007/n18/4i.html

This search shows that most cited articles use same KDD99 or DARPA dataset.

• Google Scholar Search

This search shows Darpa and KDD99 as most relevant results

• Google Search