Dataset & Hyperparameters
Two clusters with slight overlap at the border.
Epoch
0
w₁ = 0.00
w₂ = 0.00
b = 0.00
∇w₁ = 0.00 ∇w₂ = 0.00
w₂ = 0.00
b = 0.00
∇w₁ = 0.00 ∇w₂ = 0.00
Decision Boundary (Feature Space)
Class 0
Class 1
── Boundary (P=0.5)
- - P=0.25, 0.75
Training Metrics Over Time
Log-Loss
Accuracy (%)
Model: z = w₁x₁ + w₂x₂ + b
Sigmoid: σ(z) = 1/(1+e−z)
Loss: −[y·log(ŷ) + (1−y)·log(1−ŷ)]
Update: w ← w − α·∇L
Boundary: where σ(z) = 0.5 ⇒ z = 0