tff.learning.optimizers.build_sgdm

Returns a tff.learning.optimizers.Optimizer for momentum SGD.

tff.learning.optimizers.build_sgdm(
    learning_rate: optimizer.Float = 0.01,
    momentum: Optional[optimizer.Float] = None
) -> tff.learning.optimizers.Optimizer

Used in the notebooks

Used in the tutorials
Federated Reconstruction for Matrix Factorization Federated Learning for Image Classification Composing Learning Algorithms Differential Privacy in TFF TFF simulations with accelerators

This class supports the simple gradient descent and its variant with momentum.

If momentum is not used, the update rule given learning rate lr, weights w and gradients g is:

w = w - lr * g

If momentum m (a float between 0.0 and 1.0) is used, the update rule is

v = m * v + g
w = w - lr * v

where v is the velocity from previous steps of the optimizer.

Args
learning_rate	A positive float for learning rate, default to 0.01.
momentum	An optional float between 0.0 and 1.0. If None, no momentum is used.