from torch.utils.data import DataLoader
from .adadps import attach_adadps
from .dataloader import LotDataLoader, PoissonSampler
from .dpoptimizer import attach_dpoptimizer
[docs]class DPSGDManager:
"""
Manager class for privatizing DPSGD (Differentially Private Stochastic Gradient Descent) optimization.
Args:
accountant: Privacy accountant providing privacy guarantees.
optimizer_cls: Class of the optimizer to be privatized.
l2_norm_clip (float): L2 norm clip parameter for gradient clipping.
dataset: Dataset used for training.
lot_size (int): Size of the lot (local update).
batch_size (int): Size of the batch used for training.
iterations (int): Number of iterations.
smoothing (bool, optional): Whether to enable smoothing. Defaults to False.
smoothing_radius (float, optional): Smoothing radius. Defaults to 10.0.
"""
def __init__(
self,
accountant,
optimizer_cls,
l2_norm_clip,
dataset,
lot_size,
batch_size,
iterations,
smoothing=False,
smoothing_radius=10.0,
):
self.accountant = accountant
self.optimizer_cls = optimizer_cls
self.l2_norm_clip = l2_norm_clip
self.dataset = dataset
self.lot_size = lot_size
self.batch_size = batch_size
self.iterations = iterations
self.smoothing = smoothing
self.smoothing_radius = smoothing_radius
[docs] def privatize(self, noise_multiplier):
"""
Privatizes the optimizer.
Args:
noise_multiplier (float): Noise multiplier for privacy.
Returns:
tuple: Tuple containing the privatized optimizer class, lot loader function, and batch loader function.
"""
dpoptimizer_class = attach_dpoptimizer(
self.optimizer_cls,
self.accountant,
self.l2_norm_clip,
noise_multiplier,
self.lot_size,
self.batch_size,
len(self.dataset),
self.smoothing,
self.smoothing_radius,
)
def lot_loader(dp_optimizer):
return LotDataLoader(
dp_optimizer,
self.dataset,
batch_sampler=PoissonSampler(
self.dataset, self.lot_size, self.iterations
),
)
def batch_loader(lot):
return DataLoader(lot, batch_size=self.batch_size, drop_last=True)
return dpoptimizer_class, lot_loader, batch_loader
[docs]class AdaDPSManager:
"""
Manager class for privatizing AdaDPS (Adaptive Differentially Private Stochastic Gradient Descent) optimization.
Args:
accountant: Privacy accountant providing privacy guarantees.
optimizer_cls: Class of the optimizer to be privatized.
l2_norm_clip (float): L2 norm clip parameter for gradient clipping.
dataset: Dataset used for training.
lot_size (int): Size of the lot (local update).
batch_size (int): Size of the batch used for training.
iterations (int): Number of iterations.
mode (str, optional): Mode of optimization (rmsprop or adam). Defaults to "rmsprop".
beta (float, optional): Beta parameter for optimization. Defaults to 0.9.
eps_to_avoid_nan (float, optional): Epsilon parameter to avoid NaN during optimization. Defaults to 1e-8.
"""
def __init__(
self,
accountant,
optimizer_cls,
l2_norm_clip,
dataset,
lot_size,
batch_size,
iterations,
mode="rmsprop",
beta=0.9,
eps_to_avoid_nan=1e-8,
):
self.accountant = accountant
self.optimizer_cls = optimizer_cls
self.l2_norm_clip = l2_norm_clip
self.dataset = dataset
self.lot_size = lot_size
self.batch_size = batch_size
self.iterations = iterations
self.mode = mode
self.beta = beta
self.eps_to_avoid_nan = eps_to_avoid_nan
[docs] def privatize(self, noise_multiplier):
"""
Privatizes the optimizer.
Args:
noise_multiplier (float): Noise multiplier for privacy.
Returns:
tuple: Tuple containing the privatized optimizer class, lot loader function, and batch loader function.
"""
dpoptimizer_class = attach_adadps(
self.optimizer_cls,
self.accountant,
self.l2_norm_clip,
noise_multiplier,
self.lot_size,
self.batch_size,
len(self.dataset),
self.mode,
self.beta,
self.eps_to_avoid_nan,
)
def lot_loader(dp_optimizer):
return LotDataLoader(
dp_optimizer,
self.dataset,
batch_sampler=PoissonSampler(
self.dataset, self.lot_size, self.iterations
),
)
def batch_loader(lot):
return DataLoader(lot, batch_size=self.batch_size, drop_last=True)
return dpoptimizer_class, lot_loader, batch_loader
