Password Detection Neural Network

This repository contains a simple neural network implemented in Python for detecting password-like strings. The goal of this project is to identify whether a given string is similar to a password based on certain characteristics and patterns commonly found in passwords.

Results

Model file: models/rnn_model.h5

Model Performances

Precision: 0.9682656053623796
Recall: 0.9356340451371319

Example

Passwords were selected from: https://github.com/danielmiessler/SecLists/blob/master/Passwords/2020-200_most_used_passwords.txt

OriginalWord	IsPassword	Prediction
Hello	False	False
at least 8 character long	False	False
SELECT * FROM table_name;	False	False
</html>	False	False
col_name	False	False
field=	False	False
147258	True	False
123456789a	True	True
blink182	True	True
michael1	True	True
1qaz2wsx	True	True
sunshine	True	False

Usage

from detect_password import *

detector = PasswordDetector(model_path='models/rnn_model.h5')

strings = ['Hello', 'at least 8 character long', '123456789a', 'blink182']

predictions = detector.detect_passwords(strings)
for string, pred in zip(strings, predictions):
    print(f"{string}, Is Password: {pred}")

Output

Hello, Is Password: False
at least 8 character long, Is Password: False
123456789a, Is Password: True
blink182, Is Password: True

For more examples, refer to example.ipynb.

Training

To get details about model training, refer to the train_model.ipynb. The training data used for traning is extracted from the following projects, You need to download the project source code and use the get_words_from_files(dir_name) function to extract strings for training:

• Wordpress: https://github.com/WordPress/WordPress
• CCXT: https://github.com/ccxt/ccxt

The passwords used during training can be found at: https://github.com/danielmiessler/SecLists/tree/master/Passwords