diff --git a/src/fedclass.py b/src/fedclass.py
index d859bdee3793b46ea288ca35031e264e973ec053..4a1a3476ccbb86309ea64fdb796dfe0b3ff07c55 100644
--- a/src/fedclass.py
+++ b/src/fedclass.py
@@ -12,7 +12,7 @@ class Client:
"""Initialize the Client object
- Args:
+ Arguments:
id : int
unique client identifier
data : dict
@@ -70,7 +70,7 @@ class Server:
def __init__(self,model,num_clusters: int=None):
"""Initialize a Server object with an empty dictionary of cluster_models
- Args:
+ Arguments:
model: nn.Module
The nn learing model the server is associated with
num_clusters: int
diff --git a/src/metrics.py b/src/metrics.py
index eadbd06214a1078f3ae001f7a5324cc0b9682b0f..651b3240611c4fcec1cd1125213484d916a3ead3 100644
--- a/src/metrics.py
+++ b/src/metrics.py
@@ -3,7 +3,7 @@ def calc_global_metrics(labels_true: list, labels_pred: list) -> dict:
""" Calculate global metrics based on model weights
- Args:
+ Arguments:
labels_true : list
list of ground truth labels
labels_pred : list
diff --git a/src/models.py b/src/models.py
index 54fe9aa3ba461a4369770f1148aea6831a4614d6..54d82f132f36dd65d40ae40b2975ee2ca867ed0b 100644
--- a/src/models.py
+++ b/src/models.py
@@ -3,13 +3,13 @@ import torch.nn as nn
import torch.nn.functional as F
-class SimpleLinear(nn.Module):
+class SimpleLinear2(nn.Module):
""" Fully connected neural network with a single hidden layer of default size 200 and ReLU activations"""
def __init__(self, h1=200):
""" Initialization function
- Args:
+ Arguments:
h1: int
Desired size of the hidden layer
"""
@@ -21,11 +21,12 @@ class SimpleLinear(nn.Module):
""" Forward pass function through the network
- Args:
+ Arguments:
x : torch.Tensor
input image of size 28 x 28
- Returns: log_softmax probabilities of the output layer
+ Returns:
+ log_softmax probabilities of the output layer
"""
x = x.view(-1, 28 * 28)
@@ -34,33 +35,33 @@ class SimpleLinear(nn.Module):
return F.log_softmax(x, dim=1)
-class SimpleConv(nn.Module):
+class SimpleLinear(nn.Module):
""" Convolutional neural network with 3 convolutional layers and one fully connected layer
"""
- def __init__(self):
+ def __init__(self, in_size, n_channels):
""" Initialization function
"""
- super(SimpleConv, self).__init__()
- # convolutional layer
- self.conv1 = nn.Conv2d(3, 16, 3, padding=1)
- self.conv2 = nn.Conv2d(16, 32, 3, padding = 1)
- self.conv3 = nn.Conv2d(32, 16, 3, padding = 1)
- # max pooling layer
- self.pool = nn.MaxPool2d(2, 2)
+ super(SimpleLinear, self).__init__()
+
+ self.conv1 = nn.Conv2d(n_channels, 16, 3, padding=1)
+ self.conv2 = nn.Conv2d(16, 32, 3, padding=1)
+ self.conv3 = nn.Conv2d(32, 16, 3, padding=1)
- # Fully connected layer
- self.fc1 = nn.Linear(16 * 4 * 4, 10)
+ self.img_final_size = int(in_size / 8)
- # Dropout
+ self.fc1 = nn.Linear(16 * self.img_final_size * self.img_final_size, 10)
+
+ self.pool = nn.MaxPool2d(2, 2)
+
self.dropout = nn.Dropout(p=0.2)
def flatten(self, x : torch.Tensor):
"""Function to flatten a layer
- Args:
+ Arguments:
x : torch.Tensor
Returns:
@@ -72,7 +73,7 @@ class SimpleConv(nn.Module):
def forward(self, x : torch.Tensor):
""" Forward pass through the network which returns the softmax probabilities of the output layer
- Args:
+ Arguments:
x : torch.Tensor
input image to use for training
"""
diff --git a/src/utils_data.py b/src/utils_data.py
index eccfd8049c43838b4a65c67762f21a05bbe4fabd..0935676903a97fe6756798025e0432ec9b11c0d2 100644
--- a/src/utils_data.py
+++ b/src/utils_data.py
@@ -7,7 +7,7 @@ def shuffle_list(list_samples : int, seed : int) -> list:
"""Function to shuffle the samples list
- Args:
+ Arguments:
list_samples : A list of samples to shuffle
seed : Randomization seed for reproducible results
@@ -29,13 +29,12 @@ def shuffle_list(list_samples : int, seed : int) -> list:
-def create_label_dict(dataset : dict, seed : int) -> dict:
+def create_label_dict(dataset : dict) -> dict:
""" Create a dictionary of dataset samples
- Args:
+ Arguments:
dataset: The name of the dataset to use ('fashion-mnist', 'mnist', or 'kmnist')
- seed : Randomization seed for reproducible results
Returns:
A dictionary of data of the form {'x': [], 'y': []}
@@ -60,6 +59,12 @@ def create_label_dict(dataset : dict, seed : int) -> dict:
elif dataset == 'mnist':
mnist = torchvision.datasets.MNIST("datasets", download=True)
(x_train, y_train) = mnist.data, mnist.targets
+ x_train = x_train.unsqueeze(1)
+
+ elif dataset == "cifar10":
+ cifar10 = torchvision.datasets.CIFAR10("datasets", download=True)
+ (x_train, y_train) = cifar10.data, cifar10.targets
+ x_train = np.transpose(x_train, (0, 3, 1, 2))
elif dataset == 'kmnist':
(x_train, y_train), _ = kmnist.load_data()
@@ -71,21 +76,18 @@ def create_label_dict(dataset : dict, seed : int) -> dict:
for label in range(10):
- label_indices = np.where(y_train == label)[0]
-
+ label_indices = np.where(np.array(y_train) == label)[0]
label_samples_x = x_train[label_indices]
-
label_dict[label] = label_samples_x
return label_dict
-
def get_clients_data(num_clients : int, num_samples_by_label : int, dataset : dict, seed : int) -> dict:
"""Distribute a dataset evenly accross num_clients clients. Works with datasets with 10 labels
- Args:
+ Arguments:
num_clients : Number of clients of interest
num_samples_by_label : Number of samples of each labels by client
@@ -96,7 +98,7 @@ def get_clients_data(num_clients : int, num_samples_by_label : int, dataset : di
import numpy as np
- label_dict = create_label_dict(dataset, seed)
+ label_dict = create_label_dict(dataset)
clients_dictionary = {}
client_dataset = {}
@@ -125,7 +127,7 @@ def rotate_images(client: Client, rotation: int) -> None:
""" Rotate a Client's images, used for ``concept shift on features''
- Args:
+ Arguments:
client : A Client object whose dataset images we want to rotate
rotation : the rotation angle to apply 0 < angle < 360
"""
@@ -154,7 +156,7 @@ def data_preparation(client : Client, row_exp : dict) -> None:
"""Saves Dataloaders of train and test data in the Client attributes
- Args:
+ Arguments:
client : The client object to modify
row_exp : The current experiment's global parameters
"""
@@ -200,9 +202,9 @@ def get_dataset_heterogeneities(heterogeneity_type: str) -> dict:
"""
Retrieves the "skew" and "ratio" attributes of a given heterogeneity type
- Args:
+ Arguments:
heterogeneity_type : The label of the heterogeneity scenario (labels-distribution-skew, concept-shift-on-labels, quantity-skew)
- Returns
+ Returns:
A dictionary of the form {<het>: []} where <het> is the applicable heterogeneity type
"""
dict_params = {}
@@ -226,7 +228,7 @@ def setup_experiment(row_exp: dict) -> Tuple[Server, list]:
"""
Setup function to create and personalize client's data
- Args:
+ Arguments:
row_exp : The current experiment's global parameters
Returns:
@@ -237,13 +239,16 @@ def setup_experiment(row_exp: dict) -> Tuple[Server, list]:
"""
from src.models import SimpleLinear
+ from src.utils_fed import init_server_cluster
import torch
list_clients = []
torch.manual_seed(row_exp['seed'])
- model_server = Server(SimpleLinear())
+ imgs_params = {'mnist': (24,1) , 'fashion-mnist': (24,1), 'kmnist': (24,1), 'cifar10': (32,3)}
+
+ model_server = Server(SimpleLinear(in_size=imgs_params[row_exp['dataset']][0], n_channels=imgs_params[row_exp['dataset']][1]))
dict_clients = get_clients_data(row_exp['num_clients'],
row_exp['num_samples_by_label'],
@@ -255,7 +260,10 @@ def setup_experiment(row_exp: dict) -> Tuple[Server, list]:
list_clients.append(Client(i, dict_clients[i]))
list_clients = add_clients_heterogeneity(list_clients, row_exp)
-
+
+ if row_exp['exp_type'] == "client":
+ init_server_cluster(model_server, list_clients, row_exp, imgs_params['dataset'])
+
return model_server, list_clients
@@ -263,7 +271,7 @@ def setup_experiment(row_exp: dict) -> Tuple[Server, list]:
def add_clients_heterogeneity(list_clients: list, row_exp: dict) -> list:
""" Utility function to apply the relevant heterogeneity classes to each client
- Args:
+ Arguments:
list_clients : List of Client Objects with specific heterogeneity_class
row_exp : The current experiment's global parameters
Returns:
@@ -296,7 +304,7 @@ def apply_label_swap(list_clients : list, row_exp : dict, list_swaps : list) ->
""" Utility function to apply label swaps on Client images
- Args:
+ Arguments:
list_clients : List of Client Objects with specific heterogeneity_class
row_exp : The current experiment's global parameters
list_swap : List containing the labels to swap by heterogeneity class
@@ -334,7 +342,7 @@ def apply_rotation(list_clients : list, row_exp : dict) -> list:
""" Utility function to apply rotation 0,90,180 and 270 to 1/4 of Clients
- Args:
+ Arguments:
list_clients : List of Client Objects with specific heterogeneity_class
row_exp : The current experiment's global parameters
@@ -373,7 +381,7 @@ def apply_labels_skew(list_clients : list, row_exp : dict, list_skews : list, li
""" Utility function to apply label skew to Clients' data
- Args:
+ Arguments:
list_clients : List of Client Objects with specific heterogeneity_class
row_exp : The current experiment's global parameters
@@ -413,7 +421,7 @@ def apply_quantity_skew(list_clients : list, row_exp : dict, list_skews : list)
For each element in list_skews, apply the skew to an equal subset of Clients
- Args:
+ Arguments:
list_clients : List of Client Objects with specific heterogeneity_class
row_exp : The current experiment's global parameters
list_skew : List of float 0 < i < 1 with quantity skews to subsample data
@@ -455,7 +463,7 @@ def apply_features_skew(list_clients : list, row_exp : dict) -> list :
""" Utility function to apply features skew to Clients' data
- Args:
+ Arguments:
list_clients : List of Client Objects with specific heterogeneity_class
row_exp : The current experiment's global parameters
@@ -500,7 +508,7 @@ def swap_labels(labels : list, client : Client, heterogeneity_class : int) -> Cl
""" Utility Function for label swapping used for concept shift on labels. Sets the attribute "heterogeneity class"
- Args:
+ Arguments:
labels : Labels to swap
client : The Client object whose data we want to apply the swap on
Returns:
@@ -526,7 +534,7 @@ def swap_labels(labels : list, client : Client, heterogeneity_class : int) -> Cl
def centralize_data(list_clients : list) -> Tuple[DataLoader, DataLoader]:
"""Centralize data of the federated learning setup for central model comparison
- Args:
+ Arguments:
list_clients : The list of Client Objects
Returns:
@@ -564,7 +572,7 @@ def unbalancing(client : Client ,labels_list : list ,ratio_list: list) -> Client
""" Downsample the dataset of a client with each elements of the labels_list will be downsampled by the corresponding ration of ratio_list
- Args:
+ Arguments:
client : Client whose dataset we want to downsample
labels_list : Labels to downsample in the Client's dataset
ratio_list : Ratios to use for downsampling the labels
@@ -572,6 +580,7 @@ def unbalancing(client : Client ,labels_list : list ,ratio_list: list) -> Client
import pandas as pd
from imblearn.datasets import make_imbalance
+ from math import prod
def ratio_func(y, multiplier, minority_class):
@@ -584,10 +593,10 @@ def unbalancing(client : Client ,labels_list : list ,ratio_list: list) -> Client
x_train = client.data['x']
y_train = client.data['y']
- (_, i_dim,j_dim) = x_train.shape
+ orig_shape = x_train.shape
# flatten the images
- X_resampled = x_train.reshape(-1, i_dim * j_dim)
+ X_resampled = x_train.reshape(-1, prod(orig_shape[1:]))
y_resampled = y_train
for i in range(len(labels_list)):
@@ -599,7 +608,7 @@ def unbalancing(client : Client ,labels_list : list ,ratio_list: list) -> Client
sampling_strategy=ratio_func,
**{"multiplier": ratio_list[i], "minority_class": labels_list[i]})
- client.data['x'] = X_resampled.to_numpy().reshape(-1, i_dim, j_dim)
+ client.data['x'] = X_resampled.to_numpy().reshape(-1, *orig_shape[1:])
client.data['y'] = y_resampled
return client
@@ -611,7 +620,7 @@ def dilate_images(x_train : ndarray, kernel_size : tuple = (3, 3)) -> ndarray:
Make image 'bolder' for features distribution skew setup
- Args:
+ Arguments:
x_train : Input batch of images (3D array with shape (n, height, width)).
kernel_size : Size of the structuring element/kernel for dilation.
@@ -643,7 +652,7 @@ def erode_images(x_train : ndarray, kernel_size : tuple =(3, 3)) -> ndarray:
Perform erosion operation on a batch of images using a given kernel.
Make image 'finner' for features distribution skew setup
- Args:
+ Arguments:
x_train : Input batch of images (3D array with shape (n, height, width)).
kernel_size : Size of the structuring element/kernel for erosion.
@@ -675,7 +684,7 @@ def save_results(model_server : Server, row_exp : dict ) -> None:
"""
Saves model_server in row_exp['output'] as *.pth object
- Args:
+ Arguments:
model_server : The nn.Module to save
row_exp : The current experiment's global parameters
"""
diff --git a/src/utils_fed.py b/src/utils_fed.py
index 3b98f0771f1334f579b880db9b01701beaa4ea3b..7544e235ae1785d13d1e6d6e318648dbea79ecf6 100644
--- a/src/utils_fed.py
+++ b/src/utils_fed.py
@@ -7,7 +7,7 @@ def send_server_model_to_client(list_clients : list, my_server : Server) -> None
""" Function to copy the Server model to client attributes in a FL protocol
- Args:
+ Arguments:
list_clients : List of Client objects on which to set the parameter `model'
my_server : Server object with the model to copy
"""
@@ -23,7 +23,7 @@ def send_server_model_to_client(list_clients : list, my_server : Server) -> None
def send_cluster_models_to_clients(list_clients : list , my_server : Server) -> None:
""" Function to copy Server modelm to clients based on attribute client.cluster_id
- Args:
+ Arguments:
list_clients : List of Clients to update
my_server : Server from which to fetch models
"""
@@ -43,7 +43,7 @@ def model_avg(list_clients : list) -> nn.Module:
""" Utility function for the fed_avg function which creates a new model
with weights set to the weighted average of
- Args:
+ Arguments:
list_clients : List of Client whose models we want to use to perform the weighted average
Returns:
@@ -66,8 +66,7 @@ def model_avg(list_clients : list) -> nn.Module:
data_size = len(client.data_loader['train'].dataset)
- weight = data_size / total_data_size
-
+ weight = data_size / total_data_size
weighted_avg_param += client.model.state_dict()[name] * weight
param.data = weighted_avg_param #TODO: make more explicit
@@ -81,22 +80,20 @@ def fedavg(my_server : Server, list_clients : list) -> None:
The code modifies the cluster models `my_server.cluster_models[i]'
- Args:
+ Arguments:
my_server : Server model which contains the cluster models
list_clients: List of clients, each containing a PyTorch model and a data loader.
"""
if my_server.num_clusters == None:
- # Initialize a new model
+
my_server.model = model_avg(list_clients)
else :
for cluster_id in range(my_server.num_clusters):
-
- # Filter clients belonging to the current cluster
-
+
cluster_clients_list = [client for client in list_clients if client.cluster_id == cluster_id]
if len(cluster_clients_list)>0 :
@@ -109,11 +106,11 @@ def model_weight_matrix(list_clients : list) -> pd.DataFrame:
""" Create a weight matrix DataFrame using the weights of local federated models for use in the server-side CFL
- Args :
+ Arguments:
- list_clients: List of Clients with respective models
+ list_clients: List of Clients with respective models
- Returns
+ Returns:
DataFrame with weights of each model as rows
"""
@@ -124,13 +121,11 @@ def model_weight_matrix(list_clients : list) -> pd.DataFrame:
model_dict = {client.id : client.model for client in list_clients}
shapes = [param.data.numpy().shape for param in next(iter(model_dict.values())).parameters()]
-
weight_matrix_np = np.empty((len(model_dict), sum(np.prod(shape) for shape in shapes)))
for idx, (_, model) in enumerate(model_dict.items()):
model_weights = np.concatenate([param.data.numpy().flatten() for param in model.parameters()])
-
weight_matrix_np[idx, :] = model_weights
weight_matrix = pd.DataFrame(weight_matrix_np, columns=[f'w_{i+1}' for i in range(weight_matrix_np.shape[1])])
@@ -154,7 +149,6 @@ def k_means_cluster_id(weight_matrix : pd.DataFrame, k : int, seed : int) -> pd.
from sklearn.cluster import KMeans
kmeans = KMeans(n_clusters=k, random_state=seed)
-
kmeans.fit(weight_matrix)
weight_matrix['cluster'] = kmeans.labels_
@@ -168,7 +162,7 @@ def k_means_clustering(list_clients : list, num_clusters : int, seed : int) -> N
""" Performs a k-mean clustering and sets the cluser_id attribute to clients based on the result
- Args:
+ Arguments:
list_clients : List of Clients on which to perform clustering
num_clusters : Parameter to set the number of clusters needed
seed : Random seed to allow reproducibility
@@ -187,12 +181,12 @@ def k_means_clustering(list_clients : list, num_clusters : int, seed : int) -> N
-def init_server_cluster(my_server : Server, list_clients : list, row_exp : dict, p_expert_opinion : float = 0) -> None:
+def init_server_cluster(my_server : Server, list_clients : list, row_exp : dict, imgs_params: dict, p_expert_opinion : float = 0) -> None:
""" Function to initialize cluster membership for client-side CFL (sets param cluster id)
using a given distribution or completely at random.
- Args:
+ Arguments:
my_server : Server model containing one model per cluster
list_clients : List of Clients whose model we want to initialize
@@ -217,8 +211,8 @@ 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: SimpleLinear(h1=200) for cluster_id in range(row_exp['num_clusters'])}
+
+ my_server.clusters_models = {cluster_id: SimpleLinear(in_size=imgs_params[0], n_channels=imgs_params[1]) for cluster_id in range(row_exp['num_clusters'])}
for client in list_clients:
@@ -227,7 +221,7 @@ def init_server_cluster(my_server : Server, list_clients : list, row_exp : dict,
else p_expert_opinion for x in range(row_exp['num_clusters'])]
client.cluster_id = np.random.choice(range(row_exp['num_clusters']), p = probs)
-
+
client.model = copy.deepcopy(my_server.clusters_models[client.cluster_id])
return
@@ -237,7 +231,7 @@ def loss_calculation(model : nn.modules, train_loader : DataLoader) -> float:
""" Utility function to calculate average_loss across all samples <train_loader>
- Args:
+ Arguments:
model : the input server model
@@ -274,7 +268,7 @@ def loss_calculation(model : nn.modules, train_loader : DataLoader) -> float:
def set_client_cluster(my_server : Server, list_clients : list, row_exp : dict) -> None:
""" Function to calculate cluster membership for client-side CFL (sets param cluster id)
- Args:
+ Arguments:
my_server : Server model containing one model per cluster
list_clients : List of Clients whose model we want to initialize
diff --git a/src/utils_logging.py b/src/utils_logging.py
index 37ef6e0bdf69a1b33f2eb66455002334d3bbe0cb..ba4cd9390b791277b251f0aceae972be96e590b0 100644
--- a/src/utils_logging.py
+++ b/src/utils_logging.py
@@ -16,7 +16,7 @@ def cprint(msg: str, lvl: str = "info") -> None:
"""
Print message to the console at the desired logging level.
- Args:
+ Arguments:
msg (str): Message to print.
lvl (str): Logging level between "debug", "info", "warning", "error" and "critical".
The default value is "info".
diff --git a/src/utils_results.py b/src/utils_results.py
index 6252519078b1ca11bd07da3008d2b7aa422db284..01e685518de02bb9c62029e203b019856e52fed4 100644
--- a/src/utils_results.py
+++ b/src/utils_results.py
@@ -1,7 +1,8 @@
from pandas import DataFrame
from pathlib import Path
-
+from torch import tensor
+
def save_histograms() -> None:
@@ -50,7 +51,7 @@ def append_empty_clusters(list_clusters : list) -> list:
"""
Utility function for ``get_clusters'' to handle the situation where some clusters are empty by appending the clusters ID
- Args:
+ Arguments:
list_clusters: List of clusters with clients
Returns:
@@ -71,8 +72,10 @@ def append_empty_clusters(list_clusters : list) -> list:
-def get_z_nclients(df_results, x_het, y_clust, labels_heterogeneity):
+def get_z_nclients(df_results : dict, x_het : list, y_clust : list, labels_heterogeneity : list) -> list:
+ """ Returns the number of clients associated with a given heterogeneity class for each cluster"""
+
z_nclients = [0]* len(x_het)
for i in range(len(z_nclients)):
@@ -84,11 +87,22 @@ def get_z_nclients(df_results, x_het, y_clust, labels_heterogeneity):
+def plot_img(img : tensor) -> None:
+
+ """Utility function to plot an image of any shape"""
+
+ from torchvision import transforms
+ import matplotlib.pyplot as plt
+
+ plt.imshow(transforms.ToPILImage()(img))
+
+
+
def plot_histogram_clusters(df_results: DataFrame, title : str) -> None:
""" Function to create 3D Histograms of clients to cluster assignments showing client's heterogeneity class
- Args:
+ Arguments:
df_results : DataFrame containing all parameters from the resulting csv files
diff --git a/src/utils_training.py b/src/utils_training.py
index 6efdf63dbbbf7fe4d8c81a26ed9b6405d207e5f1..91e971d7e10369ceeb1f45de38db0e832430a447 100644
--- a/src/utils_training.py
+++ b/src/utils_training.py
@@ -16,13 +16,11 @@ def run_cfl_server_side(model_server : Server, list_clients : list, row_exp : di
""" Driver function for server-side cluster FL algorithm. The algorithm personalize training by clusters obtained
from model weights (k-means).
- Args:
-
- model_server : The nn.Module to save
-
- list_clients : A list of Client Objects used as nodes in the FL protocol
- row_exp : The current experiment's global parameters
+ Arguments:
+ main_model : Type of Server model needed
+ list_clients : A list of Client Objects used as nodes in the FL protocol
+ row_exp : The current experiment's global parameters
"""
from src.utils_fed import k_means_clustering
import copy
@@ -40,40 +38,36 @@ def run_cfl_server_side(model_server : Server, list_clients : list, row_exp : di
model_server = train_federated(model_server, list_clients, row_exp, use_cluster_models = True)
- list_clients = add_clients_accuracies(model_server, list_clients)
+ for client in list_clients :
+
+ acc = test_model(model_server.clusters_models[client.cluster_id], client.data_loader['test'])
+
+ setattr(client, 'accuracy', acc)
+
df_results = pd.DataFrame.from_records([c.to_dict() for c in list_clients])
return df_results
-
-
def run_cfl_client_side(model_server : Server, list_clients : list, row_exp : dict, init_cluster=True) -> pd.DataFrame:
""" Driver function for client-side cluster FL algorithm. The algorithm personalize training by clusters obtained
from model weights (k-means).
- Args:
-
- model_server : The nn.Module to save
- list_clients : A list of Client Objects used as nodes in the FL protocol
+ Arguments:
+ main_model : Type of Server model needed
+ list_clients : A list of Client Objects used as nodes in the FL protocol
row_exp : The current experiment's global parameters
-
- init_clusters : boolean indicating whether cluster assignement is done before initial training
-
+ init_cluster : A boolean indicating whether to initialize cluster prior to training
"""
- from src.utils_fed import init_server_cluster, set_client_cluster, fedavg
+ from src.utils_fed import set_client_cluster, fedavg
import torch
torch.manual_seed(row_exp['seed'])
-
- if init_cluster == True :
-
- init_server_cluster(model_server, list_clients, row_exp, p_expert_opinion=0.0)
for _ in range(row_exp['federated_rounds']):
@@ -85,22 +79,25 @@ def run_cfl_client_side(model_server : Server, list_clients : list, row_exp : di
set_client_cluster(model_server, list_clients, row_exp)
- list_clients = add_clients_accuracies(model_server, list_clients)
+ for client in list_clients :
+
+ acc = test_model(model_server.clusters_models[client.cluster_id], client.data_loader['test'])
+
+ setattr(client, 'accuracy', acc)
df_results = pd.DataFrame.from_records([c.to_dict() for c in list_clients])
return df_results
-
+
def run_benchmark(main_model : nn.Module, list_clients : list, row_exp : dict) -> pd.DataFrame:
""" Benchmark function to calculate baseline FL results and ``optimal'' personalization results if clusters are known in advance
- Args:
+ Arguments:
+
main_model : Type of Server model needed
-
list_clients : A list of Client Objects used as nodes in the FL protocol
-
row_exp : The current experiment's global parameters
"""
@@ -129,7 +126,11 @@ def run_benchmark(main_model : nn.Module, list_clients : list, row_exp : dict) -
model_trained, _ = train_central(curr_model, train_loader, row_exp)
- test_benchmark(model_trained, list_clients_filtered, test_loader, row_exp)
+ global_acc = test_model(model_trained, test_loader)
+
+ for client in list_clients_filtered :
+
+ setattr(client, 'accuracy', global_acc)
case 'global-federated':
@@ -139,55 +140,27 @@ def run_benchmark(main_model : nn.Module, list_clients : list, row_exp : dict) -
_, test_loader = centralize_data(list_clients)
- test_benchmark(model_trained.model, list_clients, test_loader, row_exp)
-
- df_results = pd.DataFrame.from_records([c.to_dict() for c in list_clients])
-
- return df_results
-
-
-def test_benchmark(model_trained : nn.Module, list_clients : list, test_loader : DataLoader, row_exp : dict):
-
- """ Tests <model_trained> on test_loader (global) dataset and sets the attribute accuracy on each Client
-
- Args:
-
- list_clients : A list of Client Objects used as nodes in the FL protocol
-
- row_exp : The current experiment's global parameters
-
- main_model : Type of Server model needed
-
- training_type : a value frmo ['global-federated', 'pers-centralized']
-
- """
-
- from src.utils_training import test_model
-
- global_acc = test_model(model_trained, test_loader)
+ global_acc = test_model(model_trained.model, test_loader)
- for client in list_clients :
+ for client in list_clients :
- #client_acc = test_model(model_trained, client.data_loader['test'])*100
+ setattr(client, 'accuracy', global_acc)
- setattr(client, 'accuracy', global_acc)
+ df_results = pd.DataFrame.from_records([c.to_dict() for c in list_clients])
- return global_acc
+ return df_results
def train_federated(main_model, list_clients, row_exp, use_cluster_models = False):
"""Controler function to launch federated learning
- Args:
-
- main_model : Server model used in our experiment
-
- list_clients : A list of Client Objects used as nodes in the FL protocol
-
- row_exp : The current experiment's global parameters
+ Arguments:
- use_cluster_models : Boolean to determine whether to use personalization by clustering
+ main_model: Server model used in our experiment
+ list_clients: A list of Client Objects used as nodes in the FL protocol
+ row_exp: The current experiment's global parameters
+ use_cluster_models: Boolean to determine whether to use personalization by clustering
"""
from src.utils_fed import send_server_model_to_client, send_cluster_models_to_clients, fedavg
@@ -220,12 +193,10 @@ def train_central(main_model, train_loader, row_exp):
""" Main training function for centralized learning
- Args:
+ Arguments:
main_model : Server model used in our experiment
-
train_loader : DataLoader with the dataset to use for training
-
row_exp : The current experiment's global parameters
"""
@@ -271,10 +242,8 @@ def test_model(model : nn.Module, test_loader : DataLoader) -> float:
""" Calcualtes model accuracy (percentage) on the <test_loader> Dataset
- Args:
-
+ Arguments:
model : the input server model
-
test_loader : DataLoader with the dataset to use for testing
"""
@@ -307,24 +276,4 @@ def test_model(model : nn.Module, test_loader : DataLoader) -> float:
accuracy = (correct / total) * 100
- return accuracy
-
-
-def add_clients_accuracies(model_server : nn.Module, list_clients : list) -> list:
-
- """
- Evaluates the cluster's models saved in <model_server> on the relevant list of clients and sets the attribute accuracy.
-
- Args:
- model_server : Server object which contains the cluster models
-
- list_clients : list of Client objects which belong to the different clusters<
- """
-
- for client in list_clients :
-
- acc = test_model(model_server.clusters_models[client.cluster_id], client.data_loader['test'])
-
- setattr(client, 'accuracy', acc)
-
- return list_clients
+ return accuracy
\ No newline at end of file