View source on GitHub |
DP subclass of tf.keras.Sequential
.
tf_privacy.DPSequential(
l2_norm_clip,
noise_multiplier,
num_microbatches=None,
use_xla=True,
layer_registry=None,
*args,
**kwargs
)
This can be used as a differentially private replacement for tf.keras.Sequential. This class implements DP-SGD using the standard Gaussian mechanism.
This class also utilizes a faster gradient clipping algorithm if the
following two conditions hold:
(i) the trainable layers of the model are keys in the input
layer_registry
,
(ii) the loss tf.Tensor
for a given batch of examples is either a
scalar or a 2D tf.Tensor
that has only one column
(i.e., tf.shape(loss)[1] == 1)
and whose i-th row corresponds to
the loss of the i-th example.
This clipping algorithm specifically computes clipped gradients at the
per-example or per microbatch (when num_microbatches
is not None)
level using the layer registry functions in layer_registry
(see
clip_grads.py for more information about the algorithm).
When instantiating this class, you need to supply several
DP-related arguments followed by the standard arguments for
Sequential
.
Examples:
# Create Model instance.
model = DPSequential(l2_norm_clip=1.0, noise_multiplier=0.5, use_xla=True,
<standard arguments>)
You should use your DPSequential instance with a standard instance
of tf.keras.Optimizer
as the optimizer, and a standard reduced loss.
You do not need to use a differentially private optimizer.
# Use a standard (non-DP) optimizer.
optimizer = tf.keras.optimizers.SGD(learning_rate=0.01)
# Use a standard reduced loss.
loss = tf.keras.losses.MeanSquaredError()
model.compile(optimizer=optimizer, loss=loss)
model.fit(train_data, train_labels, epochs=1, batch_size=32)
Methods
add
add(
layer
)
Adds a layer instance on top of the layer stack.
Args | |
---|---|
layer
|
layer instance. |
Raises | |
---|---|
TypeError
|
If layer is not a layer instance.
|
ValueError
|
In case the layer argument does not
know its input shape.
|
ValueError
|
In case the layer argument has
multiple output tensors, or is already connected
somewhere else (forbidden in Sequential models).
|
pop
pop()
Removes the last layer in the model.
Raises | |
---|---|
TypeError
|
if there are no layers in the model. |