| Definition | A Gymnasium-based RL environment simulating HTTPS Intrusion Detection to teach threat assessment and alert decision-making. |
| No of environments | 1 |
Cymnasium is a Gymnasium-based reinforcement learning environment designed to simulate an HTTPS Intrusion Detection System (HIDS). The goal is to teach an agent to decide when to signal an alert or allow nominal web server operations in response to potential malicious attacks. Using reinforcement learning, the system provides feedback to help HIDS identify and respond to different types of web attacks, such as SQL injections and PHP script calls, based on observed server logs.
The simulation is modeled on a Monte Carlo RL approach, where different types of attacks are represented as observations, and actions are either to allow a benign visit or signal a potential threat. The environment is built using the Farama Foundation’s Gymnasium API, although some basic constructs still remain in place due to time constraints. The environment includes a reset function that prepares the server, while the step function evaluates agent actions, rewarding the agent for correct threat assessments or penalizing it for errors. The environment is registered with a custom ID (“HIDS-v0”) for simulation.
This project served as an introduction to Gymnasium and reinforcement learning, highlighting the challenges of using the API and implementing a basic environment. With further development, the project could be expanded to improve HIDS’s decision-making capabilities in real-world cyber threats.