From 17c83ab4b7e58c3d30dafd628b34c6474c071d06 Mon Sep 17 00:00:00 2001
From: leahcimali <michaelbenalipro@gmail.com>
Date: Mon, 28 Oct 2024 18:02:00 +0100
Subject: [PATCH] Correct problem with linear model
---
src/models.py | 7 +-
src/utils_data.py | 223 ++++++++++++++++++++++------------------------
src/utils_fed.py | 6 +-
3 files changed, 117 insertions(+), 119 deletions(-)
diff --git a/src/models.py b/src/models.py
index cd01cbc..fe8e3fe 100644
--- a/src/models.py
+++ b/src/models.py
@@ -37,7 +37,7 @@ class ImageClassificationBase(nn.Module):
epoch, result['train_loss'], result['val_loss'], result['val_acc']))
class GenericLinearModel(ImageClassificationBase):
- def __init__(self, in_size, n_channels):
+ def __init__(self, in_size):
super().__init__()
self.in_size = in_size
self.network = nn.Sequential(
@@ -91,4 +91,7 @@ class GenericConvModel(ImageClassificationBase):
)
def forward(self, xb):
- return self.network(xb)
\ No newline at end of file
+ x = self.network[:-3](xb) # Pass through the Conv layers
+ self.img_final_size = x.shape[-1] # Get dynamic size after Conv layers
+ x = self.network[-3:](x.view(x.size(0), -1)) # Flatten and pass through FC layers
+ return x
\ No newline at end of file
diff --git a/src/utils_data.py b/src/utils_data.py
index 1d67a06..d0133dc 100644
--- a/src/utils_data.py
+++ b/src/utils_data.py
@@ -48,25 +48,34 @@ def create_label_dict(dataset : str, nn_model : str) -> dict:
import torchvision
import torchvision.transforms as transforms
+
+
if dataset == "fashion-mnist":
- fashion_mnist = torchvision.datasets.MNIST("datasets", download=True)
+ fashion_mnist = torchvision.datasets.FashionMNIST("datasets", download=True)
(x_data, y_data) = fashion_mnist.data, fashion_mnist.targets
- x_data = x_data.unsqueeze(3)
+ if nn_model == "convolutional":
+ x_data = x_data.unsqueeze(1) # Change shape to (samples, 1, H, W)
+
elif dataset == 'mnist':
mnist = torchvision.datasets.MNIST("datasets", download=True)
(x_data, y_data) = mnist.data, mnist.targets
- x_data = x_data.unsqueeze(3)
- elif dataset == "cifar10":
- cifar10 = torchvision.datasets.CIFAR10("datasets", download=True)
- (x_data, y_data) = cifar10.data, cifar10.targets
-
+ if nn_model == "convolutional":
+ x_data = x_data.unsqueeze(1) # Change shape to (samples, 1, H, W)
+
elif dataset == 'kmnist':
kmnist = torchvision.datasets.KMNIST("datasets", download=True)
- x_data = kmnist.data # This gives you the images
- x_data = x_data.unsqueeze(3)
- y_data = kmnist.targets # This gives you the labels
-
+ x_data, y_data = kmnist.data, kmnist.targets
+ if nn_model == "convolutional":
+ x_data = x_data.unsqueeze(1) # Change shape to (samples, 1, H, W)
+
+ elif dataset == "cifar10":
+ if nn_model == "linear":
+ raise ValueError("CIFAR-10 cannot be used with a linear model. Please use a convolutional model.")
+
+ cifar10 = torchvision.datasets.CIFAR10("datasets", download=True)
+ x_data, y_data = cifar10.data, cifar10.targets
+ x_data = np.transpose(x_data, (0, 3, 1, 2)) # Change shape to (samples, C, H, W)
else:
sys.exit("Unrecognized dataset. Please make sure you are using one of the following ['mnist', fashion-mnist', 'kmnist']")
@@ -155,59 +164,82 @@ from sklearn.model_selection import train_test_split
from torch.utils.data import DataLoader, TensorDataset, Dataset
import torchvision.transforms as transforms
-
-class AddGaussianNoise(object):
-
- def __init__(self, mean=0., std=1.):
- self.std = std
- self.mean = mean
-
- def __call__(self, tensor):
- import torch
- return tensor + torch.randn(tensor.size()) * self.std + self.mean
-
- def __repr__(self):
- return self.__class__.__name__ + '(mean={0}, std={1})'.format(self.mean, self.std)
-
-class AddRandomJitter(object):
-
- def __init__(self, brightness =0.5, contrast = 1, saturation = 0.1, hue = 0.5):
- self.brightness = brightness,
- self.contrast = contrast,
- self.saturation = saturation,
- self.hue = hue
-
- def __call__(self, tensor):
- import torchvision.transforms as transforms
- transform = transforms.ColorJitter(brightness = self.brightness, contrast= self.contrast,
- saturation = self.saturation, hue = self.hue)
- return transform(tensor)
-
class CustomDataset(Dataset):
-
- def __init__(self, data, labels, transform=None):
- # Ensure data is in (N, H, W, C) format
- self.data = data # Assume data is in (N, H, W, C) format
- self.labels = labels
+ """Custom dataset to apply transformations."""
+ def __init__(self, x_data, y_data, transform=None):
+ self.x_data = x_data
+ self.y_data = y_data
self.transform = transform
def __len__(self):
- return len(self.data)
+ return len(self.x_data)
def __getitem__(self, idx):
- image = self.data[idx] # Shape (H, W, C)
- label = self.labels[idx]
+ sample = self.x_data[idx]
+ label = self.y_data[idx]
- # Convert image to tensor and permute to (C, H, W)
- image = torch.tensor(image, dtype=torch.float) # Convert to tensor
- image = image.permute(2, 0, 1) # Change to (C, H, W)
-
- # Apply transformation if provided
if self.transform:
- image = self.transform(image)
-
- return image, label
+ sample = self.transform(sample)
+ return sample, label
+
+def data_transformation(row_exp : dict)-> tuple:
+ '''
+ Create transformation relative to the dataset and nn_model type
+ Arguments:
+ row_exp : The current experiment's global parameters
+ Returns:
+ tuple: A tuple containing three elements:
+ - train_transform (transforms.Compose): Transformations applied
+ to the training dataset.
+ - val_transform (transforms.Compose): Transformations applied
+ to the validation dataset.
+ - test_transform (transforms.Compose): Transformations applied
+ to the test dataset.
+ '''
+ if row_exp['dataset'] == 'cifar10':
+ train_transform = transforms.Compose([
+ transforms.RandomHorizontalFlip(),
+ transforms.RandomRotation(20),
+ transforms.RandomCrop(32, padding=4),
+ transforms.ToTensor(), # Convert to tensor
+ transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))
+ ])
+ val_transform = transforms.Compose([
+ transforms.ToTensor(),
+ transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))
+ ])
+ test_transform = transforms.Compose([
+ transforms.ToTensor(),
+ transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))
+ ])
+ elif row_exp['nn_model'] == 'convolutional': #dataset of mnist type
+ # For other datasets
+ train_transform = transforms.Compose([
+ transforms.ToTensor(),
+ transforms.Normalize((0.5,), (0.5,))
+ ])
+ val_transform = transforms.Compose([
+ transforms.ToTensor(),
+ transforms.Normalize((0.5,), (0.5,))
+ ])
+ test_transform = transforms.Compose([
+ transforms.ToTensor(),
+ transforms.Normalize((0.5,), (0.5,))
+ ])
+ else : #linear + mnist type dataset
+ # For other datasets
+ train_transform = transforms.Compose([
+ transforms.ToTensor()
+ ])
+ val_transform = transforms.Compose([
+ transforms.ToTensor()
+ ])
+ test_transform = transforms.Compose([
+ transforms.ToTensor()
+ ])
+ return train_transform, val_transform, test_transform
+
def data_preparation(client: Client, row_exp: dict) -> None:
"""Saves Dataloaders of train and test data in the Client attributes
@@ -215,57 +247,35 @@ def data_preparation(client: Client, row_exp: dict) -> None:
client : The client object to modify
row_exp : The current experiment's global parameters
"""
-
- def to_device_tensor(data, device, data_dtype):
- data = torch.tensor(data, dtype=data_dtype)
- data = data.to(device)
- return data
import torch
from sklearn.model_selection import train_test_split
from torch.utils.data import DataLoader, TensorDataset, Dataset
import torchvision.transforms as transforms
import numpy as np # Import NumPy for transpose operation
-
- # Define data augmentation transforms
- if row_exp['dataset'] == 'cifar10':
- train_transform = transforms.Compose([
- transforms.RandomHorizontalFlip(),
- transforms.RandomRotation(20), # Normalize if needed
- transforms.RandomCrop(32, padding=4),
- transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))
- ])
- # Transform for validation and test data (no augmentation, just normalization)
- test_val_transform = transforms.Compose([
- transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5)), # Normalize if needed
- ])
- else :
- train_transform = transforms.Compose([
- transforms.Normalize((0.5,), (0.5,)), # Normalize if needed
- ])
-
- # Transform for validation and test data (no augmentation, just normalization)
- test_val_transform = transforms.Compose([
- transforms.Normalize((0.5,), (0.5,)), # Normalize if needed
- ])
-
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
# Split into train, validation, and test sets
- x_data, x_test, y_data, y_test = train_test_split(client.data['x'], client.data['y'], test_size=0.3, random_state=row_exp['seed'], stratify=client.data['y'])
- x_train, x_val, y_train, y_val = train_test_split(x_data, y_data, test_size=0.25, random_state=42)
-
+ x_data, x_test, y_data, y_test = train_test_split(
+ client.data['x'], client.data['y'], test_size=0.3,
+ random_state=row_exp['seed'], stratify=client.data['y']
+ )
+ x_train, x_val, y_train, y_val = train_test_split(
+ x_data, y_data, test_size=0.25, random_state=42
+ )
+
+ # Define data augmentation transforms
+ train_transform, val_transform, test_transform = data_transformation(row_exp)
# Create datasets with transformations
train_dataset = CustomDataset(x_train, y_train, transform=train_transform)
- val_dataset = CustomDataset(x_val, y_val, transform=test_val_transform)
- test_dataset = CustomDataset(x_test, y_test, transform=test_val_transform)
+ val_dataset = CustomDataset(x_val, y_val, transform=val_transform)
+ test_dataset = CustomDataset(x_test, y_test, transform=test_transform)
# Create DataLoaders
train_loader = DataLoader(train_dataset, batch_size=128, shuffle=True)
- validation_loader = DataLoader(val_dataset, batch_size=128, shuffle=True)
- test_loader = DataLoader(test_dataset, batch_size=128, shuffle=True)
-
+ validation_loader = DataLoader(val_dataset, batch_size=128, shuffle=False)
+ test_loader = DataLoader(test_dataset, batch_size=128, shuffle=False)
# Store DataLoaders in the client object
setattr(client, 'data_loader', {'train': train_loader, 'val': validation_loader, 'test': test_loader})
setattr(client, 'train_test', {'x_train': x_train, 'x_val': x_val, 'x_test': x_test, 'y_train': y_train, 'y_val': y_val, 'y_test': y_test})
@@ -327,7 +337,7 @@ def setup_experiment(row_exp: dict) -> Tuple[Server, list]:
if row_exp['nn_model'] == "linear":
- model_server = Server(GenericLinearModel(in_size=imgs_params[row_exp['dataset']][0], n_channels=imgs_params[row_exp['dataset']][1]))
+ model_server = Server(GenericLinearModel(in_size=imgs_params[row_exp['dataset']][0]))
elif row_exp['nn_model'] == "convolutional":
@@ -633,28 +643,9 @@ def centralize_data(list_clients: list, row_exp: dict) -> Tuple[DataLoader, Data
from torch.utils.data import DataLoader,TensorDataset
import numpy as np
- if row_exp['dataset'] == 'cifar10':
- train_transform = transforms.Compose([
- transforms.RandomHorizontalFlip(),
- transforms.RandomRotation(20), # Normalize if needed
- transforms.RandomCrop(32, padding=4),
- transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))
- ])
- # Transform for validation and test data (no augmentation, just normalization)
- test_val_transform = transforms.Compose([
- transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5)), ])
- else :
- train_transform = transforms.Compose([
- transforms.Normalize((0.5,), (0.5,)), # Normalize if needed
- ])
- # Transform for validation and test data (no augmentation, just normalization)
- test_val_transform = transforms.Compose([
- transforms.Normalize((0.5,), (0.5,)), # Normalize if needed
- ])
-
+# Define data augmentation transforms
+ train_transform, test_val_transform = data_transformation(row_exp)
- device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
-
# Concatenate training data from all clients
x_train = np.concatenate([list_clients[id].train_test['x_train'] for id in range(len(list_clients))], axis=0)
y_train = np.concatenate([list_clients[id].train_test['y_train'] for id in range(len(list_clients))], axis=0)
@@ -674,9 +665,9 @@ def centralize_data(list_clients: list, row_exp: dict) -> Tuple[DataLoader, Data
# Create DataLoaders
train_loader = DataLoader(train_dataset, batch_size=128, shuffle=True)
- val_loader = DataLoader(val_dataset, batch_size=128, shuffle=False) # Validation typically not shuffled
- test_loader = DataLoader(test_dataset, batch_size=128, shuffle=False) # Test data typically not shuffled
-
+ val_loader = DataLoader(val_dataset, batch_size=128, shuffle=False)
+ test_loader = DataLoader(test_dataset, batch_size=128, shuffle=False)
+
return train_loader, val_loader, test_loader
diff --git a/src/utils_fed.py b/src/utils_fed.py
index 5c08df9..f32bcdd 100644
--- a/src/utils_fed.py
+++ b/src/utils_fed.py
@@ -214,7 +214,11 @@ def init_server_cluster(my_server : Server, list_clients : list, row_exp : dict,
p_rest = (1 - p_expert_opinion) / (row_exp['num_clusters'] - 1)
my_server.num_clusters = row_exp['num_clusters']
- my_server.clusters_models = {cluster_id: GenericConvModel(in_size=imgs_params[0], n_channels=imgs_params[1]) for cluster_id in range(row_exp['num_clusters'])}
+ if row_exp['nn_model'] == 'linear':
+ my_server.clusters_models = {cluster_id: GenericLinearModel(in_size=imgs_params[0]) for cluster_id in range(row_exp['num_clusters'])}
+
+ elif row_exp['nn_model'] == 'convolutional':
+ my_server.clusters_models = {cluster_id: GenericConvModel(in_size=imgs_params[0], n_channels=imgs_params[1]) for cluster_id in range(row_exp['num_clusters'])}
for client in list_clients:
--
GitLab