search

tpanum / towards-adversarial-phishing-detection

Implementation for publication "Towards Adversarial Phishing Detection" by Panum et al. [USENIX CSET '20]
1 stars 3 forks source link

readme

Toward Adversarial Phishing Detection

This repository contains the implementation for the publication "Towards Adversarial Phishing Detection" that is presented at USENIX CSET '20 and is designed to be used with the conda package manager.

Prepare the environment

Prior to running the code, ensure you have a conda environment with the required dependencies.

conda env create -n <myenv> -f environment.yaml
conda activate <myenv>

Models and weights

Weights for the WhiteNet is located in assets/model-weights/whitenet.pt (no adv. training) and assets/model-weights/whitenet-adv.pt (with adv. training).

These weighs were trained on a Tesla V100 GPU, using the train.py script and a data set of 37K websites across 2.5K domains.

HSL Perturbation

If you wish to experiment with HSL Perturbation, run plot.py with your desired specification.