remove last file after completion

results/1_10_fraction_retrain_res.csv

@@ -1,31 +1,6 @@
 ,accuracys,unlearning time
-0,0.7917981072555205,0.5968606472015381
-1,0.7913866410643259,1.492163896560669
-2,0.7911123302701961,2.313441276550293
-3,0.790152242490742,3.5358822345733643
-4,0.7894664655054177,4.624298810958862
-5,0.789054999314223,5.555612802505493
-6,0.7893293101083527,6.441725492477417
-7,0.7891921547112879,7.237414121627808
-8,0.7900150870936772,7.994604587554932
-9,0.7891921547112879,8.738701343536377
-10,0.7893293101083527,9.42024278640747
-11,0.7887806885200933,10.210513353347778
-12,0.7887806885200933,10.857799291610718
-13,0.7886435331230284,11.536596059799194
-14,0.7875462899465094,12.223093509674072
-15,0.786860512961185,12.898322582244873
-16,0.7869976683582499,13.568557262420654
-17,0.7863118913729256,14.259593963623047
-18,0.7861747359758606,14.833626985549927
-19,0.7863118913729256,15.40062165260315
-20,0.7861747359758606,16.014375925064087
-21,0.7859004251817309,16.656851768493652
-22,0.7854889589905363,17.292553186416626
-23,0.7843917158140172,17.955029726028442
-24,0.7843917158140172,18.59359383583069
-25,0.7827458510492388,19.34312343597412
-26,0.7830201618433685,19.972756385803223
-27,0.7823343848580442,20.579628229141235
-28,0.782471540255109,21.19595170021057
-29,0.7828830064463037,21.754724979400635
+0,0.7905637086819366,0.4226658344268799
+1,0.7908380194760664,1.123835563659668
+2,0.7896036209024825,1.758284330368042
+3,0.790152242490742,2.376178741455078
+4,0.7908380194760664,2.973546266555786

results/1_15_fraction_retrain_res.csv

@@ -1,31 +1,6 @@
 ,accuracys,unlearning time
-0,0.7761623919901248,0.6061851978302002
-1,0.7758880811959951,1.3753862380981445
-2,0.7762995473871896,1.8982110023498535
-3,0.7760252365930599,2.446392297744751
-4,0.7756137704018653,2.9644880294799805
-5,0.7762995473871896,3.443305015563965
-6,0.7758880811959951,4.151177644729614
-7,0.7762995473871896,5.0940022468566895
-8,0.7762995473871896,6.27827525138855
-9,0.7756137704018653,7.280501365661621
-10,0.7754766150048005,8.042450904846191
-11,0.7746536826224112,8.541795015335083
-12,0.7738307502400219,8.994153261184692
-13,0.7735564394458921,9.514929294586182
-14,0.7732821286517625,10.034508228302002
-15,0.772733507063503,10.539336442947388
-16,0.7721848854752434,11.148948669433594
-17,0.7709504869016596,11.707820177078247
-18,0.7716362638869839,12.151366233825684
-19,0.7710876422987245,12.578791618347168
-20,0.7720477300781786,13.05399203300476
-21,0.7717734192840489,13.489069700241089
-22,0.7709504869016596,13.999388933181763
-23,0.7699903991222055,14.516704797744751
-24,0.7713619530928542,14.988868951797485
-25,0.7710876422987245,15.51879358291626
-26,0.7709504869016596,16.006236791610718
-27,0.7709504869016596,16.543922901153564
-28,0.7708133315045947,17.11853051185608
-29,0.770539020710465,17.55792784690857
+0,0.7757509257989302,0.3392763137817383
+1,0.7757509257989302,0.7557206153869629
+2,0.7760252365930599,1.2300143241882324
+3,0.7756137704018653,1.663421869277954
+4,0.7760252365930599,2.1615872383117676

results/1_5_fraction_retrain_res.csv

@@ -1,31 +1,6 @@
 ,accuracys,unlearning time
-0,0.8138801261829653,0.8139553070068359
-1,0.8140172815800302,2.0597801208496094
-2,0.8140172815800302,3.2409777641296387
-3,0.814154436977095,4.287345886230469
-4,0.8138801261829653,5.769192218780518
-5,0.8137429707859004,7.035276651382446
-6,0.8133315045947058,8.203376054763794
-7,0.8129200384035111,9.42108964920044
-8,0.812097106021122,10.634456872940063
-9,0.812097106021122,11.921549797058105
-10,0.811959950624057,13.060882091522217
-11,0.812097106021122,14.205500364303589
-12,0.8114113290357976,15.408828020095825
-13,0.8109998628446029,16.755947828292847
-14,0.8107255520504731,17.76923155784607
-15,0.8112741736387327,18.93325114250183
-16,0.8115484844328624,20.157156467437744
-17,0.8114113290357976,21.263965129852295
-18,0.811959950624057,22.489388942718506
-19,0.8111370182416678,23.638585329055786
-20,0.812097106021122,24.870457410812378
-21,0.8114113290357976,25.924683570861816
-22,0.8108627074475381,27.062446117401123
-23,0.8112741736387327,28.360248565673828
-24,0.8111370182416678,29.612057209014893
-25,0.8109998628446029,30.784350156784058
-26,0.8103140858592786,32.129491090774536
-27,0.8108627074475381,33.685486793518066
-28,0.8104512412563435,34.84719896316528
-29,0.8103140858592786,35.942861795425415
+0,0.8134686599917706,0.7565746307373047
+1,0.8133315045947058,1.9375522136688232
+2,0.8138801261829653,3.00258207321167
+3,0.8136058153888356,4.032928705215454
+4,0.8126457276093815,5.04685378074646

results/fullretrain_res.csv

@@ -1,31 +1,6 @@
 ,accuracys,unlearning time
-0,0.8948018104512413,3.153808832168579
-1,0.8942531888629818,8.047770500183105
-2,0.8941160334659168,13.15370488166809
-3,0.8941160334659168,18.238775491714478
-4,0.8943903442600466,23.903899669647217
-5,0.8946646550541764,29.161866426467896
-6,0.8946646550541764,34.56741642951965
-7,0.893978878068852,39.76766085624695
-8,0.8948018104512413,44.90149736404419
-9,0.8950761212453711,50.2125461101532
-10,0.8946646550541764,55.36747717857361
-11,0.8950761212453711,60.54462742805481
-12,0.8954875874365656,65.36863350868225
-13,0.8949389658483061,70.5425980091095
-14,0.8946646550541764,75.46760416030884
-15,0.8943903442600466,80.56883215904236
-16,0.8937045672747223,85.88708853721619
-17,0.8934302564805925,90.91791129112244
-18,0.8927444794952681,95.90959644317627
-19,0.8927444794952681,100.96271085739136
-20,0.8923330133040736,106.17920732498169
-21,0.8924701687011384,111.32231950759888
-22,0.892881634892333,116.37431216239929
-23,0.892881634892333,121.65272212028503
-24,0.8926073240982033,126.83410930633545
-25,0.8924701687011384,132.2915494441986
-26,0.8923330133040736,137.9828565120697
-27,0.8926073240982033,142.94215846061707
-28,0.8920587025099438,148.37671399116516
-29,0.8923330133040736,153.94352960586548
+0,0.8950761212453711,3.1790289878845215
+1,0.8948018104512413,8.141436576843262
+2,0.8943903442600466,12.757618188858032
+3,0.8953504320395007,17.17648935317993
+4,0.8950761212453711,21.688454389572144

results/shards10_unlearning_res.csv

@@ -1,31 +1,6 @@
 ,accuracys,unlearning time
-0,0.8090258668134287,2.2001426219940186
-1,0.8090258668134287,2.6331143379211426
-2,0.8090258668134287,3.018749237060547
-3,0.8090258668134287,3.484102725982666
-4,0.8088882773802972,3.9745006561279297
-5,0.8088882773802972,4.306251764297485
-6,0.8091634562465603,4.689649343490601
-7,0.8091634562465603,5.156806707382202
-8,0.8090258668134287,5.631770849227905
-9,0.8093010456796919,5.97024393081665
-10,0.8091634562465603,6.455369710922241
-11,0.8091634562465603,6.812710762023926
-12,0.8091634562465603,7.247638940811157
-13,0.8093010456796919,7.7624828815460205
-14,0.8093010456796919,8.321550130844116
-15,0.8090258668134287,8.807941198348999
-16,0.8090258668134287,9.404858112335205
-17,0.8090258668134287,9.820172786712646
-18,0.8090258668134287,10.349631786346436
-19,0.8090258668134287,10.846659660339355
-20,0.8090258668134287,11.30949330329895
-21,0.8090258668134287,11.767237424850464
-22,0.8093010456796919,12.255142211914062
-23,0.8091634562465603,12.742095708847046
-24,0.8086130985140341,13.260343551635742
-25,0.8090258668134287,13.772712230682373
-26,0.8090258668134287,14.269036293029785
-27,0.8091634562465603,14.751636981964111
-28,0.8091634562465603,15.272844791412354
-29,0.8091634562465603,15.651252269744873
+0,0.8090258668134287,2.012634515762329
+1,0.8090258668134287,2.316821813583374
+2,0.8090258668134287,2.7696096897125244
+3,0.8090258668134287,3.1795125007629395
+4,0.8090258668134287,3.5385525226593018

results/shards15_unlearning_res.csv

@@ -1,31 +1,6 @@
 ,accuracys,unlearning time
-0,0.7963676389653275,2.5766398906707764
-1,0.7963676389653275,2.925642728805542
-2,0.7963676389653275,3.2963685989379883
-3,0.7963676389653275,3.575709104537964
-4,0.7963676389653275,3.900413751602173
-5,0.7960924600990644,4.219345331192017
-6,0.7960924600990644,4.55573034286499
-7,0.7960924600990644,4.827460050582886
-8,0.7960924600990644,5.1063268184661865
-9,0.7960924600990644,5.443089246749878
-10,0.7960924600990644,5.785690069198608
-11,0.7960924600990644,6.0828893184661865
-12,0.7960924600990644,6.358281850814819
-13,0.7960924600990644,6.632842063903809
-14,0.7960924600990644,6.910656213760376
-15,0.7956796917996698,7.228364706039429
-16,0.7956796917996698,7.605190753936768
-17,0.7956796917996698,7.918082237243652
-18,0.7956796917996698,8.197046995162964
-19,0.7956796917996698,8.51812481880188
-20,0.7956796917996698,8.828375816345215
-21,0.7956796917996698,9.14735460281372
-22,0.7956796917996698,9.429996013641357
-23,0.7955421023665382,9.747061252593994
-24,0.7955421023665382,10.081171989440918
-25,0.7955421023665382,10.455705404281616
-26,0.7955421023665382,10.829243421554565
-27,0.7955421023665382,11.117472648620605
-28,0.7955421023665382,11.514079570770264
-29,0.7955421023665382,11.86928677558899
+0,0.7963676389653275,2.398097038269043
+1,0.7963676389653275,2.743192434310913
+2,0.796505228398459,3.0413970947265625
+3,0.796505228398459,3.347128391265869
+4,0.796505228398459,3.6309216022491455

results/shards5_unlearning_res.csv

@@ -1,31 +1,6 @@
 ,accuracys,unlearning time
-0,0.8241607044578977,2.3396260738372803
-1,0.8242982938910292,3.370234489440918
-2,0.8244358833241607,4.0333452224731445
-3,0.8242982938910292,4.769611120223999
-4,0.8241607044578977,5.410383224487305
-5,0.8240231150247661,6.118927717208862
-6,0.8240231150247661,6.895174503326416
-7,0.8238855255916345,7.570781946182251
-8,0.8237479361585031,8.156978607177734
-9,0.8233351678591084,8.943801403045654
-10,0.8236103467253715,9.58351993560791
-11,0.82347275729224,10.199347734451294
-12,0.8237479361585031,10.879125833511353
-13,0.8237479361585031,11.650331258773804
-14,0.8240231150247661,12.463914155960083
-15,0.8241607044578977,13.124300718307495
-16,0.8241607044578977,13.843977928161621
-17,0.8242982938910292,14.789794445037842
-18,0.8244358833241607,15.582483530044556
-19,0.8244358833241607,16.335221529006958
-20,0.8247110621904238,16.953579425811768
-21,0.8247110621904238,17.663076400756836
-22,0.8249862410566868,18.315706968307495
-23,0.8249862410566868,18.94924807548523
-24,0.8251238304898184,19.57866358757019
-25,0.8249862410566868,20.244060516357422
-26,0.8248486516235554,20.938979625701904
-27,0.8252614199229499,21.571834325790405
-28,0.8252614199229499,22.25372838973999
-29,0.8252614199229499,23.12247395515442
+0,0.8241607044578977,1.821343183517456
+1,0.8241607044578977,2.3584578037261963
+2,0.8241607044578977,3.0576889514923096
+3,0.8240231150247661,3.7224462032318115
+4,0.8241607044578977,4.318850755691528

score/KNN/Cls_Fullretrain.py

@@ -3,6 +3,7 @@
 import matplotlib.pyplot as plt
 from Kbatch_KNNunlearning import KNNbase_Unlearning
 from KnnPred import knnpred
+import os
 
 
 class Fullretrain():
@@ -33,6 +34,8 @@ def unlearning(self):
             acc_temp, _ = Pred.calculate_accuracy()
             accuracys.append(acc_temp)
 
+        last_file_path = "../../data/u-unlearning{}.csv".format(next_index)
+        os.remove(last_file_path)
         res = {'accuracys': accuracys, 'unlearning time': unlearning_times}
         res_data = pd.DataFrame(res)
         res_data.to_csv('../../results/fullretrain_res.csv')

score/KNN/Cls_Knnbagging.py

@@ -4,7 +4,7 @@
 from bagging_KNNunlearning import KNNbase_Unlearning
 from KnnPred import knnpred
 import random
-
+import os
 
 class Knnbagging():
 
@@ -50,6 +50,7 @@ def unlearning(self):
             models[s].data_readin('../../data/shards_{}_{}/dataset_sharded{}.csv'.format(self.shards, self.shuffled_ordered_str, s))
             models[s].data_df.to_csv("../../data/shards_{}_{}/shard{}-unlearning0.csv".format(self.shards, self.shuffled_ordered_str, s),
                                    sep="\t", header=None, index=False)
+
         accuracys, unlearning_times, next_indexs = [], [], []
         time_start = time.time()
         for s in range(self.shards):
@@ -69,6 +70,10 @@ def unlearning(self):
             print('shards {}\'s accuracy: {}'.format(self.shards, acc_temp))
             bagging_time += time.time() - bagging_start
 
+        for s in range(self.shards):
+            last_file_path = os.path.expanduser("../../data/shards_{}_{}/shard{}-unlearning{}.csv".format(self.shards, self.shuffled_ordered_str, s, next_indexs[s]))
+            os.remove(last_file_path)
+
         res = {'accuracys': accuracys, 'unlearning time': unlearning_times}
         res_data = pd.DataFrame(res)
         res_data.to_csv('../../results/shards{}_unlearning_res.csv'.format(self.shards))
@@ -103,7 +108,7 @@ def unlearning(self):
 
 
 if __name__ == '__main__':
-    test_knnbagging = Knnbagging(10, True, 50, 5)
+    test_knnbagging = Knnbagging(10, True, 50, 3)
     test_knnbagging.unlearning()
     res = pd.read_csv('../../results/shards{}_unlearning_res.csv'.format(test_knnbagging.shards))
     accuracys = res['accuracys']

score/KNN/Cls_Sfraction.py

@@ -3,6 +3,7 @@
 import matplotlib.pyplot as plt
 from Kbatch_KNNunlearning import KNNbase_Unlearning
 from KnnPred import knnpred
+import os
 
 
 class S_fraction():
@@ -37,13 +38,15 @@ def unlearning(self):
             accuracys.append(acc_temp)
             print('accuracy: {}'.format(acc_temp))
 
+        last_file_path = os.path.expanduser('../../data/shards_{}_{}/u-unlearning{}.csv'.format(self.shards, self.shuffled_ordered_str, next_index))
+        os.remove(last_file_path)
         res = {'accuracys': accuracys, 'unlearning time': unlearning_times}
         res_data = pd.DataFrame(res)
         res_data.to_csv('../../results/1_{}_fraction_retrain_res.csv'.format(self.shards))
 
 
 if __name__ == '__main__':
-    test_s_fraction = S_fraction(10, True, 50, 5)
+    test_s_fraction = S_fraction(10, True, 50, 3)
     test_s_fraction.unlearning()
     res = pd.read_csv('../../results/1_{}_fraction_retrain_res.csv'.format(test_s_fraction.shards))
     accuracys = res['accuracys']

score/KNN/Kbatch_KNNunlearning.py

@@ -19,18 +19,19 @@ def __init__(self, shuffle=False, shards=1, remove_files_flag=True):
         self.alg, self.user_ids = None, None
         self.shuffle, self.shards = shuffle, shards
         self.remove_files_flag, self.original_flag = remove_files_flag, True
+        self.file_path = None
 
     def data_readin(self, path_to_udata):
-        file_path = os.path.expanduser(path_to_udata)
+        self.file_path = os.path.expanduser(path_to_udata)
         # print(file_path)
         # 使用Reader指定文本格式，参数line_format指定特征（列名），参数sep指定分隔符
         self.reader = Reader(line_format='user item rating timestamp', sep='\t')
         # 加载数据集
-        self.data = Dataset.load_from_file(file_path, reader=self.reader)
-        self.data_df = pd.read_csv(file_path, sep='\t', header=None, names=['user', 'item', 'rating', 'timestamp'])
+        self.data = Dataset.load_from_file(self.file_path, reader=self.reader)
+        self.data_df = pd.read_csv(self.file_path, sep='\t', header=None, names=['user', 'item', 'rating', 'timestamp'])
 
         if self.remove_files_flag and self.original_flag == False:
-            os.remove(file_path)
+            os.remove(self.file_path)
         self.original_flag = False
 
         # sorted data dataframe as user id and timestamp
@@ -156,17 +157,12 @@ def recommendation_unlearning(self, forgetting_index, batchsize):
         for _ in range(batchsize):
             uid = random.choice(self.user_ids).item()
             uids.append(uid)
-        # print("----- select user uids -----")
-        # print(uids)
 
         history_index = []
         for _ in range(batchsize):
             data_user = self.getting_user_history(uids[_])
             indexarr = data_user.index.values
             history_index_temp = random.choice(indexarr).item()
-            # dest = data_user.loc[history_index_temp]
-            # movie_name = dest["item"]
-            # movie_index = self.movie_name_to_item(movie_name)
             history_index.append(history_index_temp)
 
         self.unlearning_request(history_index)