fix HFL clustering bug

src/.ipynb_checkpoints/federated-hierarchical_v1_twoclusters-checkpoint.ipynb

@@ -53,7 +53,7 @@
     "parser.add_argument('--epochs', type=int, default=5, help=\"number of rounds of training(10)\")\n",
     "parser.add_argument('--num_users', type=int, default=100, help=\"number of users: K\")\n",
     "parser.add_argument('--num_classes', type=int, default=10, help=\"number of classes\")\n",
-    "parser.add_argument('--gpu', default=None, help=\"To use cuda, set to a specific GPU ID. Default set to use CPU.\")\n",
+    "parser.add_argument('--gpu', default=1, help=\"To use cuda, set to a specific GPU ID. Default set to use CPU.\")\n",
     "parser.add_argument('--frac', type=float, default=0.1, help='the fraction of clients: C')\n",
     "parser.add_argument('--local_ep', type=int, default=3, help=\"the number of local epochs: E\")\n",
     "parser.add_argument('--local_bs', type=int, default=10, help=\"local batch size: B\")\n",

src/federated-hierarchical_main.py

@@ -17,6 +17,7 @@ from options import args_parser
 from update import LocalUpdate, test_inference
 from models import MLP, CNNMnist, CNNFashion_Mnist, CNNCifar
 from utils import get_dataset, average_weights, exp_details
+import math
 
 
 # BUILD MODEL
@@ -58,7 +59,8 @@ def fl_train(args, train_dataset, cluster_global_model, cluster, usergrp, epochs
         # print(f'\n | Cluster Training Round : {epoch+1} |\n')
 
         cluster_global_model.train()
-        m = max(int(args.frac * len(cluster)), 1)
+        # m = max(int(args.frac * len(cluster)), 1)
+        m = max(int(math.ceil(args.frac * len(cluster))), 1)
         idxs_users = np.random.choice(cluster, m, replace=False)
 
 
@@ -78,7 +80,21 @@ def fl_train(args, train_dataset, cluster_global_model, cluster, usergrp, epochs
         cluster_loss_avg = sum(cluster_local_losses) / len(cluster_local_losses)
         cluster_train_loss.append(cluster_loss_avg)
 
-    return cluster_global_weights, cluster_loss_avg
+        # ============== EVAL ============== 
+        # Calculate avg training accuracy over all users at every epoch
+        list_acc, list_loss = [], []
+        cluster_global_model.eval()
+        for c in range(len(cluster)):
+            local_model = LocalUpdate(args=args, dataset=train_dataset,
+                                      idxs=user_groups[c], logger=logger)
+            acc, loss = local_model.inference(model=global_model)
+            list_acc.append(acc)
+            list_loss.append(loss)
+        train_accuracy.append(sum(list_acc)/len(list_acc))
+        # Add
+    print("Cluster accuracy: ", 100*train_accuracy[-1]) 
+
+    return cluster_global_model, cluster_global_weights, cluster_loss_avg
     
 
 
@@ -113,14 +129,14 @@ if __name__ == '__main__':
     B1 = np.arange(cluster_size, cluster_size+cluster_size, dtype=int)
     user_groupsB = {k:user_groups[k] for k in B1 if k in user_groups}
     print("Size of cluster 2: ", len(user_groupsB))
-    # Cluster 3
-    C1 = np.arange(2*cluster_size, 3*cluster_size, dtype=int)
-    user_groupsC = {k:user_groups[k] for k in C1 if k in user_groups}
-    print("Size of cluster 3: ", len(user_groupsC))
-    # Cluster 4
-    D1 = np.arange(3*cluster_size, 4*cluster_size, dtype=int)
-    user_groupsD = {k:user_groups[k] for k in D1 if k in user_groups}
-    print("Size of cluster 4: ", len(user_groupsD))
+    # # Cluster 3
+    # C1 = np.arange(2*cluster_size, 3*cluster_size, dtype=int)
+    # user_groupsC = {k:user_groups[k] for k in C1 if k in user_groups}
+    # print("Size of cluster 3: ", len(user_groupsC))
+    # # Cluster 4
+    # D1 = np.arange(3*cluster_size, 4*cluster_size, dtype=int)
+    # user_groupsD = {k:user_groups[k] for k in D1 if k in user_groups}
+    # print("Size of cluster 4: ", len(user_groupsD))
 
     # MODEL PARAM SUMMARY
     global_model = build_model(args, train_dataset)
@@ -153,18 +169,18 @@ if __name__ == '__main__':
     cluster_modelB.train()
     # copy weights
     cluster_modelB_weights = cluster_modelB.state_dict()
-    # Cluster C
-    cluster_modelC = build_model(args, train_dataset)
-    cluster_modelC.to(device)
-    cluster_modelC.train()
-    # copy weights
-    cluster_modelC_weights = cluster_modelC.state_dict()
-    # Cluster D
-    cluster_modelD = build_model(args, train_dataset)
-    cluster_modelD.to(device)
-    cluster_modelD.train()
-    # copy weights
-    cluster_modelD_weights = cluster_modelD.state_dict()
+    # # Cluster C
+    # cluster_modelC = build_model(args, train_dataset)
+    # cluster_modelC.to(device)
+    # cluster_modelC.train()
+    # # copy weights
+    # cluster_modelC_weights = cluster_modelC.state_dict()
+    # # Cluster D
+    # cluster_modelD = build_model(args, train_dataset)
+    # cluster_modelD.to(device)
+    # cluster_modelD.train()
+    # # copy weights
+    # cluster_modelD_weights = cluster_modelD.state_dict()
 
 
     train_loss, train_accuracy = [], []
@@ -183,21 +199,23 @@ if __name__ == '__main__':
         global_model.train()
         
         # Cluster A
-        A_weights, A_losses = fl_train(args, train_dataset, cluster_modelA, A1, user_groupsA, args.Cepochs)
+        A_model, A_weights, A_losses = fl_train(args, train_dataset, cluster_modelA, A1, user_groupsA, args.Cepochs)        
         local_weights.append(copy.deepcopy(A_weights))
-        local_losses.append(copy.deepcopy(A_losses))        
+        local_losses.append(copy.deepcopy(A_losses))    
+        cluster_modelA = A_model    
         # Cluster B
-        B_weights, B_losses = fl_train(args, train_dataset, cluster_modelB, B1, user_groupsB, args.Cepochs)
+        B_model, B_weights, B_losses = fl_train(args, train_dataset, cluster_modelB, B1, user_groupsB, args.Cepochs)
         local_weights.append(copy.deepcopy(B_weights))
         local_losses.append(copy.deepcopy(B_losses))
-        # Cluster C
-        C_weights, C_losses = fl_train(args, train_dataset, cluster_modelC, C1, user_groupsC, args.Cepochs)
-        local_weights.append(copy.deepcopy(C_weights))
-        local_losses.append(copy.deepcopy(C_losses))        
-        # Cluster D
-        D_weights, D_losses = fl_train(args, train_dataset, cluster_modelD, D1, user_groupsD, args.Cepochs)
-        local_weights.append(copy.deepcopy(D_weights))
-        local_losses.append(copy.deepcopy(D_losses))
+        cluster_modelB = B_model 
+        # # Cluster C
+        # C_weights, C_losses = fl_train(args, train_dataset, cluster_modelC, C1, user_groupsC, args.Cepochs)
+        # local_weights.append(copy.deepcopy(C_weights))
+        # local_losses.append(copy.deepcopy(C_losses))        
+        # # Cluster D
+        # D_weights, D_losses = fl_train(args, train_dataset, cluster_modelD, D1, user_groupsD, args.Cepochs)
+        # local_weights.append(copy.deepcopy(D_weights))
+        # local_losses.append(copy.deepcopy(D_losses))
         
         
         # averaging global weights
@@ -242,7 +260,7 @@ if __name__ == '__main__':
     print("|---- Test Accuracy: {:.2f}%".format(100*test_acc))
 
     # Saving the objects train_loss and train_accuracy:
-    file_name = '../save/objects/{}_{}_{}_C[{}]_iid[{}]_E[{}]_B[{}].pkl'.\
+    file_name = '../save/objects/HFL_{}_{}_{}_C[{}]_iid[{}]_E[{}]_B[{}].pkl'.\
     format(args.dataset, args.model, epoch, args.frac, args.iid,
            args.local_ep, args.local_bs)
 

src/federated-hierarchical_v1_twoclusters.ipynb

@@ -53,7 +53,7 @@
     "parser.add_argument('--epochs', type=int, default=5, help=\"number of rounds of training(10)\")\n",
     "parser.add_argument('--num_users', type=int, default=100, help=\"number of users: K\")\n",
     "parser.add_argument('--num_classes', type=int, default=10, help=\"number of classes\")\n",
-    "parser.add_argument('--gpu', default=None, help=\"To use cuda, set to a specific GPU ID. Default set to use CPU.\")\n",
+    "parser.add_argument('--gpu', default=1, help=\"To use cuda, set to a specific GPU ID. Default set to use CPU.\")\n",
     "parser.add_argument('--frac', type=float, default=0.1, help='the fraction of clients: C')\n",
     "parser.add_argument('--local_ep', type=int, default=3, help=\"the number of local epochs: E\")\n",
     "parser.add_argument('--local_bs', type=int, default=10, help=\"local batch size: B\")\n",

src/federated_main.py

@@ -144,7 +144,7 @@ if __name__ == '__main__':
     print("|---- Test Accuracy: {:.2f}%".format(100*test_acc))
 
     # Saving the objects train_loss and train_accuracy:
-    file_name = '../save/objects/{}_{}_{}_C[{}]_iid[{}]_E[{}]_B[{}].pkl'.\
+    file_name = '../save/objects/FL_{}_{}_{}_C[{}]_iid[{}]_E[{}]_B[{}].pkl'.\
         format(args.dataset, args.model, args.epochs, args.frac, args.iid,
                args.local_ep, args.local_bs)