yoni-birman / spirel

Security Policy Implementation using Reinforcement Learning (SPIREL) - Cost-effective ensemble models selection using deep reinforcement learning

SPIREL

Security Policy Implementation using REinforcement Learning This code is provided as part of our work - generating a cost-effectivy security policy automatically, using reinforcement learning.

Prerequisites

The SPIREL framework is built with Python 3.6, and mainly relies on ChainerRL reinforcement learning (RL) library, as well as OpenAI Gym for the RL environment.
In the provided requirement.txt file, you can find the libraries that should be installed for our framework to work. This can be done using the following command:

pip install -r requirements.txt

Setup

Then, in the root directory (where we see the file setup.py), you should run the following command, which installs our SPIREL environment:

pip install -e .

Note that the -e flag means that we can edit the files in our installed library, without reinstalling it.

Usage

We provided an example file (example.py), which demonstrates the usage in our environment.