Assessing Password Strength with Machine Learning in Python

[timber hedge]

Hello.
In my make today, I am writing an article about password security with machine learning.
I created two different wordlists, weak and strong, for the project to work.
and it gave the results according to the input I entered.
https://denizhalil.com/2024/02/02/assessing-password-strength-with-machine-learning-in-python/

[violet fulcrum]

I’m not sure why machine learning needs to be involved in determining if a password is strong or weak. We already have these factors explicitly outlined in rules for people.

1. longer password is better because you increase the power of the combinations (ie 26^n vs 26^n+1 for single case alphabetical)
2. adding symbols and numbers and not just alphabetical will boost your password strength because the base of that exponent is increased (ie (n)^8 vs (n+1)^8).
3. keep special words and nouns out of your passwords (ie mo pet names or street names)
4. don’t use words in general. Obfuscate them if you decide to use them but the modern way of creating a password is to come up with a phrase where each word is represented in each character.

This is a good paper and you present your ideas well but I’m overall not seeing the “why” for ML in this application.

[timber hedge]

the idea here is to create an idea,
to do a study with machine security and cyber security and learn.
Of course, I know the factors you mentioned.

[violet fulcrum]

Of course, I apologize if my comments came across as offensive

[timber hedge]

of course no

[stable oracle]

It's a good introduction. You can also try looking at features like entropy or unigram distributions

[timber hedge]

thank you so much