# Applications of Machine Learning in Cyber Security 10 Points (i) Provide 3 example applications of machine learning in cybersecurity? Your examples d

CS973-HW1
Q1 Applications of Machine Learning in Cyber Security 10 Points
(i) Provide 3 example applications of machine learning in cybersecurity? Your examples do not need to go into technical details – just a sentence each about 3 problems in cyber security that can be solved using machine learning will be enough.
(ii) Among the 3 problems you selected above, choose one and explain in 2-3 brief sentences why you think machine learning based solution will work for solving the problem?
Q2 Supervised vs. Unsupervised Machine Learning
10 Points
(i) Explain in no more than 3-4 sentences the salient differences between supervised and unsupervised machine learning.
1/4
10/11/23, 6:39 PM
(ii) Give examples of two cyber security problems such that one is more easily solved by supervised machine learning and the other is more easily solved by unsupervised machine learning.
Q3 Robust Machine Learning Model
10 Points
(i) A machine learning model is trained by training data set. Further it is applied on test data set to calibrate the model for accuracy, precision etc. Once the model seems to give satisfactory performance on test data, we apply such a model on previously unseen data (data that was neither used for training or testing) and do classification or regression as the case may be. In this context, explain in 2-3 brief sentences what is meant by a robust machine learning model.
(ii) In the context of cyber security problems being solved by machine learning why is it important that the models be robust?
Q4 Classification vs. Regression Problems
10 Points
2/4
10/11/23, 6:39 PM
(i) Explain in few short sentences, the differences between classification vs. regression.
(ii) Can you think of a problem relevant to cyber security which may require regression rather than classification?
Q5 Bayesian Probability 10 Points
Suppose 10% of all IP packets entering an organization are malicious. Also, among the malicious packets, 75% of malicious packets contain a set of indicators. However, 25% of benign packets also contain those indicators. As a result, by inspecting the indicators in the packets, one cannot guarantee whether the packet is malicious or benign. Suppose you find a packet that has the set of indicators. What is the probability that it is a malicious packet? Please show your work with explanations.
Submit & View Submission
3/4
10/11/23, 6:39 PM

X
Don`t copy text!