async def compute_federated_secure_sum(
self, arg: federated_resolving_strategy.FederatedResolvingStrategyValue
) -> federated_resolving_strategy.FederatedResolvingStrategyValue:
py_typecheck.check_type(arg.internal_representation, structure.Struct)
py_typecheck.check_len(arg.internal_representation, 2)
logging.warning(
'The implementation of the `tff.federated_secure_sum` intrinsic '
'provided by the `tff.backends.test` runtime uses no cryptography.')
server_ex = self._target_executors[placement_literals.SERVER][0]
value = federated_resolving_strategy.FederatedResolvingStrategyValue(
arg.internal_representation[0], arg.type_signature[0])
bitwidth = await arg.internal_representation[1].compute()
bitwidth_type = arg.type_signature[1]
sum_result, mask, fn = await asyncio.gather(
self.compute_federated_sum(value),
executor_utils.embed_tf_scalar_constant(self._executor, bitwidth_type,
2**bitwidth - 1),
executor_utils.embed_tf_binary_operator(self._executor, bitwidth_type,
tf.math.mod))
fn_arg = await server_ex.create_struct([
sum_result.internal_representation[0],
mask.internal_representation,
])
fn_arg_type = computation_types.FederatedType(
fn_arg.type_signature, placement_literals.SERVER, all_equal=True)
arg = federated_resolving_strategy.FederatedResolvingStrategyValue(
structure.Struct([
(None, fn.internal_representation),
(None, [fn_arg]),
]), computation_types.StructType([fn.type_signature, fn_arg_type]))
return await self.compute_federated_map_all_equal(arg)
async def _compute_intrinsic_federated_sum(self, arg):
type_utils.check_federated_type(
arg.type_signature, placement=placement_literals.CLIENTS)
zero, plus, identity = tuple(await asyncio.gather(*[
executor_utils.embed_tf_scalar_constant(self, arg.type_signature.member,
0),
executor_utils.embed_tf_binary_operator(self, arg.type_signature.member,
tf.add),
self.create_value(
computation_factory.create_lambda_identity(
arg.type_signature.member),
type_factory.unary_op(arg.type_signature.member))
]))
aggregate_args = await self.create_tuple([arg, zero, plus, plus, identity])
return await self._compute_intrinsic_federated_aggregate(aggregate_args)
async def compute_federated_sum(
self,
arg: FederatedComposingStrategyValue) -> FederatedComposingStrategyValue:
type_analysis.check_federated_type(
arg.type_signature, placement=placement_literals.CLIENTS)
id_comp, id_type = tensorflow_computation_factory.create_identity(
arg.type_signature.member)
zero, plus, identity = await asyncio.gather(
executor_utils.embed_tf_scalar_constant(self._executor,
arg.type_signature.member, 0),
executor_utils.embed_tf_binary_operator(self._executor,
arg.type_signature.member,
tf.add),
self._executor.create_value(id_comp, id_type))
aggregate_args = await self._executor.create_struct(
[arg, zero, plus, plus, identity])
return await self.compute_federated_aggregate(aggregate_args)
async def _compute_intrinsic_federated_mean(self, arg):
arg_sum = await self._compute_intrinsic_federated_sum(arg)
member_type = arg_sum.type_signature.member
count = float(len(arg.internal_representation))
if count < 1.0:
raise RuntimeError('Cannot compute a federated mean over an empty group.')
child = self._target_executors[placement_literals.SERVER][0]
factor, multiply = tuple(await asyncio.gather(*[
executor_utils.embed_tf_scalar_constant(child, member_type,
float(1.0 / count)),
executor_utils.embed_tf_binary_operator(child, member_type, tf.multiply)
]))
multiply_arg = await child.create_tuple(
anonymous_tuple.AnonymousTuple([(None,
arg_sum.internal_representation[0]),
(None, factor)]))
result = await child.create_call(multiply, multiply_arg)
return FederatingExecutorValue([result], arg_sum.type_signature)
async def _compute_intrinsic_federated_sum(self, arg):
py_typecheck.check_type(arg.type_signature, computation_types.FederatedType)
zero, plus = tuple(await
asyncio.gather(*[
executor_utils.embed_tf_scalar_constant(
self, arg.type_signature.member, 0),
executor_utils.embed_tf_binary_operator(
self, arg.type_signature.member, tf.add)
]))
return await self._compute_intrinsic_federated_reduce(
FederatingExecutorValue(
anonymous_tuple.AnonymousTuple([
(None, arg.internal_representation),
(None, zero.internal_representation),
(None, plus.internal_representation)
]),
computation_types.NamedTupleType(
(arg.type_signature, zero.type_signature, plus.type_signature)))
)
async def compute_federated_sum(
self, arg: FederatedResolvingStrategyValue
) -> FederatedResolvingStrategyValue:
py_typecheck.check_type(arg.type_signature,
computation_types.FederatedType)
zero, plus = await asyncio.gather(
executor_utils.embed_tf_scalar_constant(self._executor,
arg.type_signature.member,
0),
executor_utils.embed_tf_binary_operator(self._executor,
arg.type_signature.member,
tf.add))
return await self.compute_federated_reduce(
FederatedResolvingStrategyValue(
structure.Struct([(None, arg.internal_representation),
(None, zero.internal_representation),
(None, plus.internal_representation)]),
computation_types.StructType(
(arg.type_signature, zero.type_signature,
plus.type_signature))))
async def _compute_intrinsic_federated_mean(self, arg):
member_type = arg.type_signature.member
ones = await self.create_value(
1, type_factory.at_clients(member_type, all_equal=True))
totals = (await self._compute_intrinsic_federated_sum(
await self._compute_intrinsic_federated_zip_at_clients(
await self.create_tuple([arg, ones])))).internal_representation
py_typecheck.check_type(totals, executor_value_base.ExecutorValue)
fed_sum, count = tuple(await asyncio.gather(
self._parent_executor.create_selection(totals, index=0),
self._parent_executor.create_selection(totals, index=1)))
count_val = await count.compute()
factor, multiply = tuple(await asyncio.gather(*[
executor_utils.embed_tf_scalar_constant(
self._parent_executor, member_type, float(1.0 / count_val)),
executor_utils.embed_tf_binary_operator(self._parent_executor,
member_type, tf.multiply)
]))
multiply_arg = await self._parent_executor.create_tuple([fed_sum, factor])
result = await self._parent_executor.create_call(multiply, multiply_arg)
return CompositeValue(result, type_factory.at_server(member_type))
