SEED CoG 2021 paper - "Adversarial Reinforcement Learning for Procedural Content Generation".

ea.com/seed/news/cog2021-adversarial-rl-conten…
CoG 21ea.com/seed/news/cog2021-adversarial-rl-conten…Visit site

About CoG 21

CoG 21 is a 2021 IEEE Conference on Games research paper from EA SEED that introduces an adversarial reinforcement learning approach for procedural content generation (PCG) and improving reinforcement learning (RL) agent generalization. The method uses two paired RL agents: a Generator that creates game environments, and a Solver that attempts to complete those environments, with the Generator rewarded for producing challenges that are solvable but sufficiently difficult. The approach also incorporates auxiliary inputs as control variables to increase diversity and control over generated environments, and the paper demonstrates improved generalization for RL agents compared to previous PCG methods.

Key features

  • Adversarial RL approach for procedural content generation
  • Dual RL agent system (Generator and Solver)
  • Auxiliary input control for environment generation
  • Improves RL agent generalization to new environments
  • Creates solvable, previously unseen game environments
Best for: Game developers and reinforcement learning researchers focused on procedural content generation and agent generalization.
Good for
  • Demonstrated improved RL agent generalization versus prior methods
  • Enables creation of novel, solvable game environments
  • Supports controlled, diverse environment generation via auxiliary inputs

CoG 21 at a glance

Type
Online tool
Pricing
Not confirmed. Check the site
Platforms
Web
Open source
Not stated
Language
English
Last checked
Oct 6, 2026 · Online when checked
Listed
Oct 5, 2026

CoG 21 alternatives

Tools and sites you could use instead, ranked by how closely they match.

All 7 alternatives

Mexty

mexty.ai

Mexty creates simulations, assessments, learning paths, and interactive courses, then supports delivery and tracking through its LMS.

Popular in AI Writing Tools

Well-known listings from the same topic.

All 190

CoG 21 FAQ

What is the core contribution of the CoG 21 research paper?
It introduces an adversarial reinforcement learning method for procedural content generation and improved RL agent generalization.
How does the Generator-Solver agent system function?
The Generator creates environments, the Solver tests them, and the Generator is rewarded for making solvable but challenging environments.
What feature allows control over generated environment diversity?
Auxiliary inputs are used as control variables to drive diversity and control of environment generation.
Is something wrong or out of date?

Tell us and an editor will check it.

Own this website?

Claim the listing and suggest updates yourself. An editor checks every claim first.

Claim this listing