ACE — Adaptive Content Engine

Reinforcement learning driven VR training for cybersecurity. ACE uses PPO and AIRL to adjust challenge difficulty in real time based on learner performance.

See architecture Interactive demo

What ACE Does

  • Analyzes telemetry (time, mistakes, hints, action sequences).
  • Estimates reward via AIRL; adjusts difficulty with PPO.
  • Delivers adaptive hints and pacing inside a Unity VR lab.

Why It Matters

  • Boosts engagement and skill transfer for cyber defense tasks.
  • Personalizes lab flow without manual scripting.
  • Supports safe, explainable training through logged decisions.

Architecture

VR Front-End (Unity) Telemetry / Features AIRL Reward Estimation PPO Difficulty Policy Adaptive Hints & Difficulty
Hover a node to see what it does.

Interactive Demos

Learning Curve (Simulated)

Average episodic reward over training epochs (illustrative).

Gridworld Agent (Toy)

A tiny agent moves toward a goal; speed increases when it “learns”.

Interested in the research?

This work is part of an active research effort. A public summary/paper link will be added upon release.

Contact me