How can I employ SOM for this problem

Dear all,

I want to do Unsupervised Machine Learning Using Self-Organised Map (SOM). I have manually labeled my data to either malicious user behaviour: "1" (binary) or benign user behaviour: "0" (binary) to provide a platform to evaluate the unsupervised machine learning approach. My dataset is unbalanced. This means the number of benign events are way more than the number of malicious events (33 malicious events out of 2643 total events). I am rather a beginner in SOM filed and I would like to know how can I employ SOM for my problem. I would appreciate any help .

Regards,

Ngh

Answer 1

[Greg Heath]

What you have proposed makes no sense:

1. SOM is for unsupervised learning.

2. If you know the categories then you should use supervised learning.

3. Search using

    greg newrb

Greg

[naghmeh moradpoor]

Hi Greg, Thanks for your answer. My dataset is not balanced so when I used supervised ML it gave me poor results in terms of accuracy. I am just doing some initial experiments but not sure how to interpret the output when I used SOM on my dataset. Regards, Ngh