Gradient Descent
gradient-descent
optimizer that minimizes loss by following gradient
Iteratively adjusts weights: w = w - lr * ∇L
Variants: SGD, mini-batch GD
Used to find cnn filters
optimizer that minimizes loss by following gradient
Iteratively adjusts weights: w = w - lr * ∇L
Variants: SGD, mini-batch GD
Used to find cnn filters