ML Math Lab — Signal vs Noise Survival Simulator
The glowing river is the Signal. The storm is Noise. Your AI orb must follow the pattern.
Mode: Explore
Goal: learn Signal vs Noise
Score: 0
Section 1 — What you will learn
By the end you will understand
  • Signal: the real pattern (the glowing river).
  • Noise: random mess that tries to fool you (wind + chaos).
  • Overfitting: the AI follows every firefly (memorizes noise).
  • Generalization: the AI follows the river, even during storms.
  • Outliers / Drift / Sensor noise: real-world “evil modes” of noise.
Section 2 — Controls
Challenge adds win/score + auto-stop on a “smart” model.
Noise is VISUAL here: wind lines + firefly wobble + sensor glitches.
This is the “real world expansion pack”.
Few fireflies = easy to get fooled by noise.
Think: rigid robot vs overthinking robot.
Shield helps AI ignore distractions.
How to play: click inside the arena to drop “probe pulses”. The AI orb predicts where the river is.
Yellow probes = training probes. Cyan probes = test probes (hidden until Reveal Signal).

Section 3 — Live stats
Train Error (Practice)
Test Error (Truth)
Hidden
Noise Confusion
Overfit Risk
Generalization Gap
Score
0

Section 4 — Math (simple)
What the AI is doing Live, in kid-language
Loading…
🧠 Kid Coach
Loading…
Tip: Reveal Signal = truth moment.
Section 5 — The Chaos Arena
hidden river (Signal, revealed) practice probes truth probes AI orb path error beam stabilizer shield glow
Live explanation
Move sliders → watch the world change. Noise is the storm. Signal is the river.
Game HUD
Explore: learn what noise looks like in real life.
Noise Meter (storm)
Higher noise = more chaos.
Overfit Meter (gap)
Reveal Signal to see if the AI is fooling itself.
Stability Meter (shield)
More λ = smoother behavior (less wiggle).
Quick actions
Win idea: low practice error + low truth error + small gap = learned Signal, not Noise.