Multiplication
Repeated addition or scaling.
- Commutative for scalars, NOT for matrices.
- Distributes over addition.
- Two negatives make a positive.
When to use: Linear layers, weighted sums, learning-rate scaling, dot products.
Example: Weighted sum: y = w₁·x₁ + w₂·x₂ + b. GD step: θ_new = θ − α·∇L.