ADD file via upload

Edge.py

@@ -0,0 +1,99 @@
+import numpy as np
+from models.Update import *
+from models.Fed import *
+from models.test import *
+from utils.utilis import *
+import copy
+
+def test_with_loss(net_glob, dataset_test, args):
+
+    # testing
+    acc_test, loss_test = test_img(net_glob, dataset_test, args)
+
+    print("Testing accuracy: {:.2f}".format(acc_test))
+
+    return acc_test.item(), loss_test
+
+def split_list_into_365_parts(input_list):
+    n = len(input_list)  # 获取输入列表的长度
+    # 计算每个子列表的基本长度
+    base_size = n // 365
+    # 计算需要额外添加一个元素的子列表的数量
+    remainder = n % 365
+
+    result = []
+    start = 0  # 开始索引
+
+    for i in range(365):
+        # 如果索引小于余数，当前子列表多一个元素
+        end = start + base_size + (1 if i < remainder else 0)
+        # 切分列表并添加到结果中
+        result.append(input_list[start:end])
+        start = end  # 更新下一个子列表的开始索引
+
+    return result
+
+def Edge(args, net_glob, dataset_train, dataset_test, dict_users, hosp_Model_datavolume, center_assignments):
+    net_glob.train()
+    
+    print("Start Edge")
+
+    num_hosp = args.num_centers
+    
+    isNext = True
+    today = 1
+    m = 1
+
+    ratio = [0.46, 0.31,0.12,0.11]
+    
+    # training
+    acc = []
+    GB = []
+    transf = sum(hosp_Model_datavolume)
+    total_transf = 0
+    
+    w_glob = net_glob.state_dict()
+    
+    # dict_day = [split_list_into_365_parts(dict_hosp[idx]) for idx in range(len(dict_hosp))]
+    
+    while isNext:
+        if today >= args.physical_time:
+            isNext = False
+        # print("*" * 80)
+        print("today{:3d}".format(today))
+        # day7 = list(range(today,today+7))
+        for idx_hosp in range(num_hosp):
+            w_locals = []
+            lens = []
+            net_glob.load_state_dict(w_glob)
+            l = 0
+            client_train_list = random.sample(
+                    center_assignments[idx_hosp], m
+                )
+            # for day in day7:
+            l = 0
+            for idx_client in client_train_list:
+                local = LocalUpdate(
+                        args=args, dataset=dataset_train, idxs=dict_users[idx_client]
+                    )
+                for iter in range(args.epochs):
+                    w = local.train(net=copy.deepcopy(net_glob).to(args.device))
+                    net_glob.load_state_dict(w)
+                l += len(dict_users[idx_client])
+            w = net_glob.state_dict()
+            w_locals.append(copy.deepcopy(w))
+            l *= ratio[idx_hosp]
+            lens.append(l)
+            # lens.append(len(dict_hosp[idx_hosp]))
+        today+=1
+            # update global weights
+        w_glob = Aggregation(w_locals, lens)
+        total_transf += transf
+
+        item_acc, item_loss = test_with_loss(net_glob, dataset_test, args)
+        ta, tl = test_with_loss(net_glob, dataset_train, args)
+        acc.append(item_acc)
+        GB.append(total_transf)
+
+    save_result(acc, "test_acc", args)
+    save_result(GB, "test_GB", args)