How To Use Machine Learning in Cyber Security
Machine Learning in Cyber Security – An intrusion detection system (IDS) is a type of software that works by monitoring computer systems or networks and detect any suspicious activities. The earliest form of IDS was invented in 19080 by James Anderson, which had a set of tools that administrators would use to monitor logs and other audit trails.
Over the years, IDS has technologically advanced to keep up with the ever-growing digital world. Researchers and experts have continuously updated the technology to ensure that they improve efficiency while at the same time prevent interference with network performance.
Deep learning is a subsystem of Machine learning, whose architectures can be used in multiple fields such as speech recognition, bioinformatics etc. In 1986, Rina Dechter introduced the word deep learning to the world of machine learning and in 2000, Igor Aizenberg and his counterparts used the term in artificial neural networks. The impact that deep learning has in the industry had not been felt until in the early 2000s and its applications widely began in 2010.
Dataset is a type of image recognition dataset that is used to develop algorithms for deep learning, self taught learning and unsupervised feature learning. When coming up with a project in machine learning, we have to use a dataset that will be used to train the model that you come up with on how they can carry out any necessary action.
One can only build capable IDS if you have a well functioning data set. When you have a data set that has a good amount of usable real time data that will help you to train the IDS and test it. One such data set is the NSL-KDD data set. This is the best data set that can be used to assess how the IDS are performing. This is because it has a lower detection time and high accuracy rate. The CIDDS-002 is also a data set that you can use to train the IDS.
It is most suitable for a small business network where there is unidirectional flow of traffic. This data set will give you a technical report with all the information concerning the data set as well as anonymized IP addresses. It is a publicly available data set and thus easy to access.
We can classify machine-learning algorithms in different ways. There is the supervised machine learning algorithms, unsupervised machine learning algorithms, semi-supervised machine learning algorithms and semi-supervised learning.
K-Means Clustering is one of the easiest and popular algorithms of Machine Learning. The variable K is used to represent the number of groups and the algorithm is thus used to find groups in the data. Using the features of the data set, this algorithm assigns data to each of the K groups. The data points are then grouped in terms of similarity.
Random Forest Classifier combines multiple algorithms to help in classification. It is able to come up with different decisions using the given data subsets. It thereafter comes up with a mean of all the votes to make a decision on the type of test.
Machine Learning Approach
Choosing the dataset
To create our project, we will use the NSL-KDD dataset. Prior to starting the project, research was undertaken on the best dataset that would assess the performance of our IDS system. NSL-KDD is a dataset that has a fast detection rate as well as a high accuracy level.
Setting up the environment
We will have to set up an environment that will be ideal for using ML and DL to improve our IDS system. A powerful computer system will be needed with a powerful CPU, high RAM of over 8GB, a great operating system as well as a GPU NVIDIA GeForce GTX 940 or higher.
Data pre-processing and creating the model
We can then process the data set to make it more readable and easier to process. The next step is to identify the best model with the highest accuracy rate. The next step will be the training phase where you will not use abnormal data, as this will be instantly be discarded by the IDS. The testing phase is the next step to help identify the performance score of our model.
Machine Learning Implementation
The last step is to create a website where we will make the project available to users. Users can access the improved IDS system on the website.
References
Dhanabal L & Shantharajah S.P (2015) A study on NSL-KDD Dataset for Intrusion Detection System Based on Classification Algorithms: International Journal of Advanced Research in Computer and Communication Engineering Vol 4 (6)
R.Sommer, V. Paxson, Outside the Closed World : On Using Machine Learning For Network Intrusion Detection, in: IEEE Symposium on Security and Privacy, IEEE, 2010, pp. 305-316.
Machine Learning Relevant Links
IT Computer Dissertation Topics
Big Data Strategy Decision Making
What other machine learning in cyber security strategies do you have for making sure your tweets last as long as possible? Let us know in the comments!”
