dd-m / homoglyph

Detecting Homoglyph Attacks with a Siamese Neural Network

This is sample code for training the siamese neural network and comparing it to edit distance based techniques.

This was tested with 1.2.2 version of keras and python 2.7. We recommend having a GPU to speed up processing.

Prereq

You will need a copy of Arial.ttf to put in the code directory before running. This file should be in the same directory of the python code. If not, you will see an error similar to this:

Traceback (most recent call last):
  File "run_siamese.py", line 144, in <module>
    X1_train = generate_imgs([x[0] for x in data['train']], font_location, font_size, image_size, text_location)
  File "run_siamese.py", line 47, in generate_imgs
    font = ImageFont.truetype(font_location, font_size)
  File "/home/<user>/anaconda2/lib/python2.7/site-packages/PIL/ImageFont.py", line 239, in truetype
    return FreeTypeFont(font, size, index, encoding)
  File "/home/<user>/anaconda2/lib/python2.7/site-packages/PIL/ImageFont.py", line 128, in __init__
    self.font = core.getfont(font, size, index, encoding)
IOError: cannot open resource

Mac

You can view this for getting Arial.ttf: https://support.apple.com/guide/font-book/install-and-validate-fonts-fntbk1000/mac

Ubuntu

You can install ttf-mscorefonts-installer

Windows

Look in c:\Windows\Fonts

Code params

The code is very simple and has a couple parameters that can be changed inline. The first is data type. For running the code on process data, set the dataset_type to "process":

dataset_type = 'process'

For domains:

dataset_type = 'domain'

The second parameter allows the code to run on a subset of the data to get results faster. Note that a smaller dataset will get worse results. To set this parameter to a subset set isFast to True:

isFast = True

Too set to the full dataset do:

isFast = False

Run code

Simply run the python code like:

python run_siamese.py