Machine Learning Decision Boundary Lab

Model & Inductive Bias MLP (6 Units)

Classification Architecture

Pedagogical Benchmark Presets

MLP Hidden Neurons (Layer 1) 6

Learning Rate (η) 0.05

Training Telemetry

Epochs 120

Loss (BCE) 0.082

Accuracy 97.5%

Samples 40

What is Machine Learning? Rather than hardcoding fixed geometric rules, machine learning adjusts parameters ($W, b$) via gradient descent to discover continuous decision manifolds that separate empirical observations.

Feature Space Manifold [x₁, x₂ ∈ ±3.0]

Class Cobalt (1)

Class Crimson (0)

Probe

Probe Coords (x₁, x₂) 0.00, 0.00

P(Class Cobalt | x) 0.52

Decision Margin +0.02

Gestures: Tap canvas to plant point • Drag points to warp manifold • Drag white ring probe to evaluate certainty.

Add: Class 1 Class 0