本文详细介绍了HS2算法在超图上的应用，通过Swendsen-Wang采样构建超图，并利用主动学习策略选择关键样本点。在去除截超边后，采用最短路径寻找关键超边，通过MSSP（最小割超边选择）方法选取无标记顶点。代码实现中包含了广度优先搜索和随机选点策略，用于在有限预算下进行样本选择。实验结果显示了HS2算法在不同预算下的运行情况。
目标文献：
I (Eli) Chien，Huozhi Zhou and Pan Li . HS2：Active learning over hypergraphs with pointwise and pairwise queries. In: Proceedings of the 22nd International Conference on Artificial Intelligence and Statistics (AISTATS) 2019, Naha,
Okinawa, Japan. 【CCF C类会议】
存在的case：
作者在文中没有交代通过MSSP找到最短路径中的中间超边后，如何从中间超边中选择关键样本点。所以代码中使用随机选点法获取中间超边中的任意一个无标记顶点。
要点：
- 超图构建使用了SWS（Swendsen-Wang Sampling）构建order=5，edgeNum=样本数的超图，这里调用matlab代码。ps：需要python3.6以下版本
- 使用了广度优先所有BFS（Breadth-first search）寻找两个不同类别已标记样本的之间的最短路径。
- 使用MSSP获取关键超边，使用Random choice获取关键超边的关键样本顶点。
写了一整天才搞出来，都成代码复现机器人了，太累了！
不懂可与我交流。包括HS2的存在不足！
本稿未提供sws超图构建的matlab代码！原因你懂！别人的东东！
import numpy as np
import matlab
import matlab.engine
from copy import deepcopy
from itertools import combinations,product
from collections import OrderedDict,deque
from sklearn.datasets import load_iris
from scipy.spatial.distance import pdist,squareform
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import accuracy_score
from sklearn.model_selection import StratifiedKFold
engine = matlab.engine.start_matlab()
class PWHS2():
def __init__(self,X,y,neiNum,Labeled,Budget,X_test,y_test):
self.X = X
self.y = y
self.labels = np.unique(y)
self.labNum = len(self.labels)
self.labeled = list(Labeled)
self.label_pair = list(combinations(self.labels, 2))
self.labeled_dict = self.init_labeled_dict()
self.N = len(y)
self.budget = deepcopy(Budget)
self.budgetLeft = deepcopy(Budget)
self.recordNote = [i for i in np.arange(self.labNum, self.budget + 1, self.labNum)]
self.initNum = len(self.labels)
self.unlabeled = self.init_unlabeled()
self.neiNum = neiNum
self.dist_matrix = squareform(pdist(X=X, metric='euclidean'))
self.X_test = X_test
self.y_test = y_test
self.nn_matrix = np.zeros((self.N,self.neiNum))
self.nn_list = []
self.nn_cost = []
self.hyper_graph = self.init_hyper_graph()
###--------------------------
self.AccList = []
self.MAEList = []
self.QYRList = []
class tree(object):
def __init__(self, edx):
self.edx = edx
self.parent = None
self.children = []
def init_unlabeled(self):
unlabeled = [_ for _ in range(self.N)]
for idx in self.labeled:
unlabeled.remove(idx)
return unlabeled
def init_labeled_dict(self):
labeled_dict = OrderedDict()
for idx in self.labeled:
labeled_dict[self.y[idx]] = []
for idx in self.labeled:
labeled_dict[self.y[idx]].append(idx)
return labeled_dict
def init_hyper_graph(self):
nn_list = []
nn_cost = []
for i in range(self.N):
ord_idx = np.argsort(self.dist_matrix[i])
neibor = []
for j in range(self.neiNum + 1):
if i != ord_idx[j]:
neibor.append(ord_idx[j])
neibor = np.array(neibor)
self.nn_matrix[i] = neibor
for i in range(self.N-1):
for j in range(i,self.N):
if i in self.nn_matrix[j] or j in self.nn_matrix[i]:
nn_list.append([i,j])
nn_cost.append(self.dist_matrix[i,j])
###调整为matlab索引：所有索引+1
for i in range(len(nn_list)):
nn_list[i][0] += 1
nn_list[i][1] += 1
nn_list = np.array(nn_list)
# print("边的个数=",len(nn_list))
nn_cost = np.array(nn_cost)
nn_cost = np.vstack(nn_cost)
e = matlab.double(nn_list.tolist())
cost = matlab.double(nn_cost.tolist())
#####调用SWS采样法构建超图################
hyper_edges = engine.sws(e, cost, self.N)
hyper_edges = np.array(hyper_edges)
hyper_edges = hyper_edges.T
####调整为python索引：所有索引减去1
for i in range(hyper_edges.shape[0]):
for j in range(hyper_edges.shape[1]):
hyper_edges[i,j] -= 1
########################################
hyper_graph = OrderedDict()
for idx in range(len(hyper_edges)):
hyper_graph[idx] = hyper_edges[idx]
return hyper_graph
def remove_cuts(self):
diff_pair_list = []
for labpair in self.label_pair:
idx_pair = product(self.labeled_dict[labpair[0]], self.labeled_dict[labpair[1]])
for pair in idx_pair:
diff_pair_list.append(pair)
### 当一个超边中有diff_pair 则删除这个边
cut_edge_idxs = []
for idx, edge in self.hyper_graph.items():
for pair in diff_pair_list:
if pair[0] in edge and pair[1] in edge:
cut_edge_idxs.append(idx)
break
else:
continue
###如果cutedge存在，开始删除cutedge
print("发现截超边CutEdges=",cut_edge_idxs)
if cut_edge_idxs:
for idx in cut_edge_idxs:
del self.hyper_graph[idx]
def find_children_edge(self,edge):
children = []
for idx in self.hyper_graph[edge]:
for edx, hyperedge in self.hyper_graph.items():
if idx in hyperedge:
children.append(edx)
return children
def BFS(self,pair):
source = pair[0]
target = pair[1]
###########获取目标顶点所在的边####################
target_visited = []
for edx, edge in self.hyper_graph.items():
if target in edge:
target_visited.append(edx)
###如果目标顶点不存在于任何边直接返回空
if len(target_visited) == 0:
return []
################################################
###########获取源头顶点所在的边###################
source_edges = []
for edx, edge in self.hyper_graph.items():
if source in edge:
source_edges.append(edx)
###如果源头顶点不存在于任何边直接返回空
if len(source_edges) == 0:
return []
####################正式开始#####################
visited = []
for edx in source_edges:
visited.append(edx)
queue = deque([])
TreeDict = OrderedDict()
target_edge = None
for edx in source_edges:
children = self.find_children_edge(edx)
queue.append([edx,children])
TreeDict[edx] = self.tree(edx)
TreeDict[edx].children = children
while queue:
parent, children = queue.popleft()
if parent in target_visited:
target_edge = parent
break
for child in children:
if child not in visited:
visited.append(child)
grandsons = self.find_children_edge(child)
queue.append((child,grandsons))
TreeDict[child] = self.tree(child)
TreeDict[child].children = grandsons
TreeDict[child].parent = parent
if target_edge == None:
return []
else:
path = []
current_edge = target_edge
while current_edge != None:
path.append(current_edge)
current_edge = TreeDict[current_edge].parent
return path
def mssp(self):
edge_len_dict = OrderedDict()
edge_path_dict = OrderedDict()
for labpair in self.label_pair:
idx_pairs = product(self.labeled_dict[labpair[0]], self.labeled_dict[labpair[1]])
for pair in idx_pairs:
edge_path_dict[pair] = self.BFS(pair)
if edge_path_dict[pair]:
edge_len_dict[pair] = len(edge_path_dict[pair])
else:
edge_len_dict[pair] = np.inf
tar_pair = min(edge_len_dict,key=edge_len_dict.get)
vertex = None
if edge_path_dict[tar_pair]:
path_jdx = int(np.floor(edge_len_dict[tar_pair]/2))
mid_edge = edge_path_dict[tar_pair][path_jdx]
mid_edge_samples = list(deepcopy(self.hyper_graph[mid_edge]))
while mid_edge_samples:
vertex = np.random.choice(mid_edge_samples,size=1,replace=False)[0]
mid_edge_samples.remove(vertex)
if vertex not in self.labeled:
break
else:
vertex = None
continue
return vertex
def pairwise_HS2(self):
self.remove_cuts()
while self.budgetLeft > 0:
print("进入S2>>>>>>>>>>>>>>>>>>>>>>>>>剩余预算{}>>>>>>>>>>>>>>>>>>>>>>>>>>".format(self.budgetLeft))
vertex = self.mssp()
if vertex == None:
vertex = np.random.choice(self.unlabeled, size=1, replace=False)[0]
print("执行了随机选点获取无标记顶点：",vertex)
else:
print("执行MSSP选点获取无标记顶点：",vertex)
vertex = int(vertex) ###强制类型转换为int
current_budget = len(self.labels)
for lab in self.labels:
if self.budgetLeft <= 0:
break
else:
self.budgetLeft -= 1
current_budget -= 1
#####测试并保存分类指标
if (self.budget - self.budgetLeft) in self.recordNote:
self.pred()
#####################
if current_budget <= 1:
self.labeled.append(vertex)
self.labeled_dict[self.y[vertex]].append(vertex)
self.unlabeled.remove(vertex)
break
else:
if self.y[vertex] == lab:
self.labeled.append(vertex)
self.labeled_dict[self.y[vertex]].append(vertex)
self.unlabeled.remove(vertex)
break
else:
continue
print("已标记列表",self.labeled)
print("已标记字典",self.labeled_dict)
self.remove_cuts()
if self.budget <= 0:
break
def pred(self):
model = LogisticRegression(solver='newton-cg', penalty='l2')
model.fit(self.X[self.labeled],self.y[self.labeled])
y_pred = model.predict(self.X_test)
self.AccList.append(accuracy_score(y_true=self.y_test,y_pred=y_pred))
self.MAEList.append(np.mean(abs(self.y_test - y_pred)))
self.QYRList.append((len(self.labeled)-self.initNum)/(self.budget-self.budgetLeft))
if __name__ == '__main__':
X, y = load_iris(return_X_y=True)
class_num = len(np.unique(y))
budget = 10 * class_num
round = 5
ACC = np.zeros(int(budget/class_num))
MAE = np.zeros(int(budget/class_num))
QYR = np.zeros(int(budget/class_num))
for t in range(round):
SKF = StratifiedKFold(n_splits=10, shuffle=True)
for train_idx, test_idx in SKF.split(X, y):
train_X = X[train_idx]
train_y = y[train_idx]
test_X = X[test_idx]
test_y = y[test_idx]
labeled = []
label_dict = OrderedDict()
for lab in np.unique(train_y):
label_dict[lab] = []
for idx in range(len(train_y)):
label_dict[train_y[idx]].append(idx)
for idxlist in label_dict.values():
for jdx in np.random.choice(idxlist,size=1,replace=False):
labeled.append(jdx)
print("标记预算",budget)
pwhs2 = PWHS2(X=train_X,y=train_y,neiNum=5,Labeled=labeled,Budget=budget,X_test=test_X,y_test=test_y)
print(pwhs2.hyper_graph)
print(len(pwhs2.hyper_graph))
pwhs2.remove_cuts()
print(pwhs2.hyper_graph)
print(len(pwhs2.hyper_graph))
print("开始执行HS2")
pwhs2.pairwise_HS2()
print("剩余标记预算",pwhs2.budgetLeft)
break
break输出：
标记预算 30
OrderedDict([(0, array([34., 1., 35., 21., 14.])), (1, array([111., 66., 69., 127., 62.])), (2, array([14., 13., 43., 29., 1.])), (3, array([ 54., 106., 86., 111., 64.])), (4, array([31., 15., 27., 32., 43.])), (5, array([20., 32., 17., 16., 37.])), (6, array([ 94., 63., 78., 131., 84.])), (7, array([ 70., 106., 126., 99., 57.])), (8, array([18., 13., 4., 26., 31.])), (9, array([ 56., 126., 120., 117., 48.])), (10, array([109., 78., 83., 75., 46.])), (11, array([ 86., 83., 134., 133., 91.])), (12, array([ 70., 132., 74., 87., 119.])), (13, array([ 86., 118., 93., 51., 130.])), (14, array([ 54., 52., 108., 64., 80.])), (15, array([101., 126., 99., 133., 102.])), (16, array([119., 85., 117., 66., 102.])), (17, array([43., 42., 20., 18., 36.])), (18, array([ 70., 82., 109., 115., 77.])), (19, array([124., 56., 58., 130., 107.])), (20, array([14., 22., 30., 18., 19.])), (21, array([27., 24., 28., 4., 11.])), (22, array([ 0., 4., 42., 13., 35.])), (23, array([ 87., 112., 45., 95., 67.])), (24, array([100., 70., 65., 104., 51.])), (25, array([ 81., 45., 127., 90., 121.])), (26, array([ 46., 93., 119., 127., 106.])), (27, array([ 78., 73., 115., 131., 63.])), (28, array([ 49., 54., 118., 84., 76.])), (29, array([ 84., 99., 131., 105., 119.])), (30, array([ 54., 110., 57., 122., 46.])), (31, array([37., 7., 8., 38., 41.])), (32, array([113., 46., 103., 63., 96.])), (33, array([ 85., 50., 103., 106., 102.])), (34, array([ 1., 12., 13., 35., 43.])), (35, array([38., 3., 21., 4., 34.])), (36, array([32., 23., 11., 36., 22.])), (37, array([ 5., 20., 25., 9., 39.])), (38, array([14., 1., 42., 22., 9.])), (39, array([29., 23., 22., 33., 20.])), (40, array([ 92., 109., 88., 52., 104.])), (41, array([38., 22., 36., 2., 13.])), (42, array([14., 15., 17., 32., 12.])), (43, array([ 50., 105., 48., 95., 121.])), (44, array([114., 56., 101., 50., 127.])), (45, array([105., 121., 82., 88., 67.])), (46, array([20., 12., 23., 32., 33.])), (47, array([ 51., 70., 107., 67., 57.])), (48, array([127., 109., 113., 75., 48.])), (49, array([ 98., 92., 113., 111., 73.])), (50, array([ 83., 111., 131., 108., 54.])), (51, array([15., 23., 29., 25., 35.])), (52, array([35., 7., 12., 14., 27.])), (53, array([ 89., 92., 64., 112., 68.])), (54, array([30., 24., 12., 41., 35.])), (55, array([35., 34., 22., 7., 1.])), (56, array([32., 27., 28., 16., 23.])), (57, array([28., 35., 2., 31., 13.])), (58, array([103., 121., 95., 123., 70.])), (59, array([34., 2., 27., 33., 32.])), (60, array([19., 8., 15., 18., 6.])), (61, array([56., 54., 51., 81., 68.])), (62, array([ 52., 100., 106., 111., 78.])), (63, array([14., 9., 24., 36., 38.])), (64, array([ 9., 22., 41., 26., 8.])), (65, array([42., 29., 32., 11., 28.])), (66, array([10., 17., 2., 41., 31.])), (67, array([108., 80., 101., 104., 119.])), (68, array([31., 34., 23., 8., 0.])), (69, array([26., 22., 16., 33., 34.])), (70, array([30., 16., 19., 43., 21.])), (71, array([ 69., 114., 98., 101., 126.])), (72, array([16., 12., 10., 7., 6.])), (73, array([32., 36., 13., 0., 34.])), (74, array([108., 78., 56., 127., 104.])), (75, array([14., 2., 7., 23., 35.])), (76, array([ 53., 62., 115., 52., 61.])), (77, array([13., 34., 12., 0., 25.])), (78, array([ 4., 43., 44., 33., 34.])), (79, array([117., 100., 73., 67., 113.])), (80, array([ 64., 48., 98., 110., 67.])), (81, array([ 9., 39., 24., 7., 20.])), (82, array([114., 85., 105., 108., 65.])), (83, array([30., 33., 0., 27., 2.])), (84, array([ 9., 27., 30., 13., 10.])), (85, array([29., 35., 44., 34., 6.])), (86, array([ 96., 56., 109., 79., 51.])), (87, array([18., 6., 5., 43., 1.])), (88, array([ 5., 23., 32., 44., 12.])), (89, array([ 53., 67., 60., 100., 61.])), (90, array([23., 36., 44., 30., 29.])), (91, array([104., 103., 50., 99., 127.])), (92, array([118., 84., 99., 50., 49.])), (93, array([44., 16., 4., 25., 19.])), (94, array([95., 50., 85., 76., 84.])), (95, array([ 99., 131., 133., 77., 123.])), (96, array([ 90., 128., 49., 62., 127.])), (97, array([103., 113., 107., 128., 65.])), (98, array([132., 107., 84., 79., 85.])), (99, array([10., 18., 15., 14., 2.])), (100, array([16., 14., 4., 28., 22.])), (101, array([123., 130., 114., 54., 100.])), (102, array([116., 118., 76., 82., 55.])), (103, array([20., 17., 44., 18., 2.])), (104, array([15., 4., 33., 35., 41.])), (105, array([18., 13., 43., 2., 15.])), (106, array([31., 13., 15., 9., 38.])), (107, array([19., 4., 2., 35., 31.])), (108, array([118., 105., 99., 57., 96.])), (109, array([ 1., 28., 13., 17., 2.])), (110, array([39., 2., 8., 36., 23.])), (111, array([56., 70., 55., 86., 89.])), (112, array([ 93., 78., 88., 113., 54.])), (113, array([39., 13., 16., 24., 18.])), (114, array([ 68., 48., 67., 117., 116.])), (115, array([12., 21., 30., 34., 27.])), (116, array([21., 35., 22., 15., 26.])), (117, array([28., 25., 30., 14., 41.])), (118, array([124., 79., 103., 127., 64.])), (119, array([ 3., 12., 15., 5., 8.])), (120, array([28., 39., 32., 22., 9.])), (121, array([ 99., 82., 50., 112., 122.])), (122, array([40., 18., 31., 30., 38.])), (123, array([43., 38., 40., 1., 4.])), (124, array([12., 13., 31., 26., 27.])), (125, array([116., 53., 119., 50., 63.])), (126, array([34., 20., 44., 31., 12.])), (127, array([130., 118., 105., 111., 76.])), (128, array([11., 20., 10., 7., 14.])), (129, array([19., 20., 27., 4., 7.])), (130, array([ 60., 57., 125., 117., 81.])), (131, array([43., 14., 22., 12., 23.])), (132, array([31., 18., 38., 37., 34.])), (133, array([16., 26., 44., 9., 1.])), (134, array([114., 89., 129., 48., 60.]))])
135
截超边= []
OrderedDict([(0, array([34., 1., 35., 21., 14.])), (1, array([111., 66., 69., 127., 62.])), (2, array([14., 13., 43., 29., 1.])), (3, array([ 54., 106., 86., 111., 64.])), (4, array([31., 15., 27., 32., 43.])), (5, array([20., 32., 17., 16., 37.])), (6, array([ 94., 63., 78., 131., 84.])), (7, array([ 70., 106., 126., 99., 57.])), (8, array([18., 13., 4., 26., 31.])), (9, array([ 56., 126., 120., 117., 48.])), (10, array([109., 78., 83., 75., 46.])), (11, array([ 86., 83., 134., 133., 91.])), (12, array([ 70., 132., 74., 87., 119.])), (13, array([ 86., 118., 93., 51., 130.])), (14, array([ 54., 52., 108., 64., 80.])), (15, array([101., 126., 99., 133., 102.])), (16, array([119., 85., 117., 66., 102.])), (17, array([43., 42., 20., 18., 36.])), (18, array([ 70., 82., 109., 115., 77.])), (19, array([124., 56., 58., 130., 107.])), (20, array([14., 22., 30., 18., 19.])), (21, array([27., 24., 28., 4., 11.])), (22, array([ 0., 4., 42., 13., 35.])), (23, array([ 87., 112., 45., 95., 67.])), (24, array([100., 70., 65., 104., 51.])), (25, array([ 81., 45., 127., 90., 121.])), (26, array([ 46., 93., 119., 127., 106.])), (27, array([ 78., 73., 115., 131., 63.])), (28, array([ 49., 54., 118., 84., 76.])), (29, array([ 84., 99., 131., 105., 119.])), (30, array([ 54., 110., 57., 122., 46.])), (31, array([37., 7., 8., 38., 41.])), (32, array([113., 46., 103., 63., 96.])), (33, array([ 85., 50., 103., 106., 102.])), (34, array([ 1., 12., 13., 35., 43.])), (35, array([38., 3., 21., 4., 34.])), (36, array([32., 23., 11., 36., 22.])), (37, array([ 5., 20., 25., 9., 39.])), (38, array([14., 1., 42., 22., 9.])), (39, array([29., 23., 22., 33., 20.])), (40, array([ 92., 109., 88., 52., 104.])), (41, array([38., 22., 36., 2., 13.])), (42, array([14., 15., 17., 32., 12.])), (43, array([ 50., 105., 48., 95., 121.])), (44, array([114., 56., 101., 50., 127.])), (45, array([105., 121., 82., 88., 67.])), (46, array([20., 12., 23., 32., 33.])), (47, array([ 51., 70., 107., 67., 57.])), (48, array([127., 109., 113., 75., 48.])), (49, array([ 98., 92., 113., 111., 73.])), (50, array([ 83., 111., 131., 108., 54.])), (51, array([15., 23., 29., 25., 35.])), (52, array([35., 7., 12., 14., 27.])), (53, array([ 89., 92., 64., 112., 68.])), (54, array([30., 24., 12., 41., 35.])), (55, array([35., 34., 22., 7., 1.])), (56, array([32., 27., 28., 16., 23.])), (57, array([28., 35., 2., 31., 13.])), (58, array([103., 121., 95., 123., 70.])), (59, array([34., 2., 27., 33., 32.])), (60, array([19., 8., 15., 18., 6.])), (61, array([56., 54., 51., 81., 68.])), (62, array([ 52., 100., 106., 111., 78.])), (63, array([14., 9., 24., 36., 38.])), (64, array([ 9., 22., 41., 26., 8.])), (65, array([42., 29., 32., 11., 28.])), (66, array([10., 17., 2., 41., 31.])), (67, array([108., 80., 101., 104., 119.])), (68, array([31., 34., 23., 8., 0.])), (69, array([26., 22., 16., 33., 34.])), (70, array([30., 16., 19., 43., 21.])), (71, array([ 69., 114., 98., 101., 126.])), (72, array([16., 12., 10., 7., 6.])), (73, array([32., 36., 13., 0., 34.])), (74, array([108., 78., 56., 127., 104.])), (75, array([14., 2., 7., 23., 35.])), (76, array([ 53., 62., 115., 52., 61.])), (77, array([13., 34., 12., 0., 25.])), (78, array([ 4., 43., 44., 33., 34.])), (79, array([117., 100., 73., 67., 113.])), (80, array([ 64., 48., 98., 110., 67.])), (81, array([ 9., 39., 24., 7., 20.])), (82, array([114., 85., 105., 108., 65.])), (83, array([30., 33., 0., 27., 2.])), (84, array([ 9., 27., 30., 13., 10.])), (85, array([29., 35., 44., 34., 6.])), (86, array([ 96., 56., 109., 79., 51.])), (87, array([18., 6., 5., 43., 1.])), (88, array([ 5., 23., 32., 44., 12.])), (89, array([ 53., 67., 60., 100., 61.])), (90, array([23., 36., 44., 30., 29.])), (91, array([104., 103., 50., 99., 127.])), (92, array([118., 84., 99., 50., 49.])), (93, array([44., 16., 4., 25., 19.])), (94, array([95., 50., 85., 76., 84.])), (95, array([ 99., 131., 133., 77., 123.])), (96, array([ 90., 128., 49., 62., 127.])), (97, array([103., 113., 107., 128., 65.])), (98, array([132., 107., 84., 79., 85.])), (99, array([10., 18., 15., 14., 2.])), (100, array([16., 14., 4., 28., 22.])), (101, array([123., 130., 114., 54., 100.])), (102, array([116., 118., 76., 82., 55.])), (103, array([20., 17., 44., 18., 2.])), (104, array([15., 4., 33., 35., 41.])), (105, array([18., 13., 43., 2., 15.])), (106, array([31., 13., 15., 9., 38.])), (107, array([19., 4., 2., 35., 31.])), (108, array([118., 105., 99., 57., 96.])), (109, array([ 1., 28., 13., 17., 2.])), (110, array([39., 2., 8., 36., 23.])), (111, array([56., 70., 55., 86., 89.])), (112, array([ 93., 78., 88., 113., 54.])), (113, array([39., 13., 16., 24., 18.])), (114, array([ 68., 48., 67., 117., 116.])), (115, array([12., 21., 30., 34., 27.])), (116, array([21., 35., 22., 15., 26.])), (117, array([28., 25., 30., 14., 41.])), (118, array([124., 79., 103., 127., 64.])), (119, array([ 3., 12., 15., 5., 8.])), (120, array([28., 39., 32., 22., 9.])), (121, array([ 99., 82., 50., 112., 122.])), (122, array([40., 18., 31., 30., 38.])), (123, array([43., 38., 40., 1., 4.])), (124, array([12., 13., 31., 26., 27.])), (125, array([116., 53., 119., 50., 63.])), (126, array([34., 20., 44., 31., 12.])), (127, array([130., 118., 105., 111., 76.])), (128, array([11., 20., 10., 7., 14.])), (129, array([19., 20., 27., 4., 7.])), (130, array([ 60., 57., 125., 117., 81.])), (131, array([43., 14., 22., 12., 23.])), (132, array([31., 18., 38., 37., 34.])), (133, array([16., 26., 44., 9., 1.])), (134, array([114., 89., 129., 48., 60.]))])
135
开始执行HS2
截超边= []
进入S2>>>>>>>>>>>>>>>>>>>>>>>>>剩余预算30>>>>>>>>>>>>>>>>>>>>>>>>>>
执行MSSP选点获取无标记顶点： 70.0
已标记列表 [1, 55, 103, 70]
已标记字典 OrderedDict([(0, [1]), (1, [55, 70]), (2, [103])])
截超边= [58]
进入S2>>>>>>>>>>>>>>>>>>>>>>>>>剩余预算28>>>>>>>>>>>>>>>>>>>>>>>>>>
执行MSSP选点获取无标记顶点： 57.0
已标记列表 [1, 55, 103, 70, 57]
已标记字典 OrderedDict([(0, [1]), (1, [55, 70, 57]), (2, [103])])
截超边= []
进入S2>>>>>>>>>>>>>>>>>>>>>>>>>剩余预算26>>>>>>>>>>>>>>>>>>>>>>>>>>
执行MSSP选点获取无标记顶点： 106.0
已标记列表 [1, 55, 103, 70, 57, 106]
已标记字典 OrderedDict([(0, [1]), (1, [55, 70, 57]), (2, [103, 106])])
截超边= [7]
进入S2>>>>>>>>>>>>>>>>>>>>>>>>>剩余预算24>>>>>>>>>>>>>>>>>>>>>>>>>>
执行MSSP选点获取无标记顶点： 89.0
已标记列表 [1, 55, 103, 70, 57, 106, 89]
已标记字典 OrderedDict([(0, [1]), (1, [55, 70, 57, 89]), (2, [103, 106])])
截超边= []
进入S2>>>>>>>>>>>>>>>>>>>>>>>>>剩余预算22>>>>>>>>>>>>>>>>>>>>>>>>>>
执行MSSP选点获取无标记顶点： 86.0
已标记列表 [1, 55, 103, 70, 57, 106, 89, 86]
已标记字典 OrderedDict([(0, [1]), (1, [55, 70, 57, 89, 86]), (2, [103, 106])])
截超边= [3]
进入S2>>>>>>>>>>>>>>>>>>>>>>>>>剩余预算20>>>>>>>>>>>>>>>>>>>>>>>>>>
执行MSSP选点获取无标记顶点： 104.0
已标记列表 [1, 55, 103, 70, 57, 106, 89, 86, 104]
已标记字典 OrderedDict([(0, [1]), (1, [55, 70, 57, 89, 86]), (2, [103, 106, 104])])
截超边= [24]
进入S2>>>>>>>>>>>>>>>>>>>>>>>>>剩余预算18>>>>>>>>>>>>>>>>>>>>>>>>>>
执行MSSP选点获取无标记顶点： 56.0
已标记列表 [1, 55, 103, 70, 57, 106, 89, 86, 104, 56]
已标记字典 OrderedDict([(0, [1]), (1, [55, 70, 57, 89, 86, 56]), (2, [103, 106, 104])])
截超边= [74]
进入S2>>>>>>>>>>>>>>>>>>>>>>>>>剩余预算16>>>>>>>>>>>>>>>>>>>>>>>>>>
执行MSSP选点获取无标记顶点： 67.0
已标记列表 [1, 55, 103, 70, 57, 106, 89, 86, 104, 56, 67]
已标记字典 OrderedDict([(0, [1]), (1, [55, 70, 57, 89, 86, 56, 67]), (2, [103, 106, 104])])
截超边= []
进入S2>>>>>>>>>>>>>>>>>>>>>>>>>剩余预算14>>>>>>>>>>>>>>>>>>>>>>>>>>
执行MSSP选点获取无标记顶点： 107.0
已标记列表 [1, 55, 103, 70, 57, 106, 89, 86, 104, 56, 67, 107]
已标记字典 OrderedDict([(0, [1]), (1, [55, 70, 57, 89, 86, 56, 67]), (2, [103, 106, 104, 107])])
截超边= [19, 47]
进入S2>>>>>>>>>>>>>>>>>>>>>>>>>剩余预算12>>>>>>>>>>>>>>>>>>>>>>>>>>
执行MSSP选点获取无标记顶点： 132.0
已标记列表 [1, 55, 103, 70, 57, 106, 89, 86, 104, 56, 67, 107, 132]
已标记字典 OrderedDict([(0, [1]), (1, [55, 70, 57, 89, 86, 56, 67]), (2, [103, 106, 104, 107, 132])])
截超边= [12]
进入S2>>>>>>>>>>>>>>>>>>>>>>>>>剩余预算10>>>>>>>>>>>>>>>>>>>>>>>>>>
执行MSSP选点获取无标记顶点： 109.0
已标记列表 [1, 55, 103, 70, 57, 106, 89, 86, 104, 56, 67, 107, 132, 109]
已标记字典 OrderedDict([(0, [1]), (1, [55, 70, 57, 89, 86, 56, 67]), (2, [103, 106, 104, 107, 132, 109])])
截超边= [18, 86]
进入S2>>>>>>>>>>>>>>>>>>>>>>>>>剩余预算8>>>>>>>>>>>>>>>>>>>>>>>>>>
执行MSSP选点获取无标记顶点： 46.0
已标记列表 [1, 55, 103, 70, 57, 106, 89, 86, 104, 56, 67, 107, 132, 109, 46]
已标记字典 OrderedDict([(0, [1]), (1, [55, 70, 57, 89, 86, 56, 67, 46]), (2, [103, 106, 104, 107, 132, 109])])
截超边= [10, 26, 32]
进入S2>>>>>>>>>>>>>>>>>>>>>>>>>剩余预算6>>>>>>>>>>>>>>>>>>>>>>>>>>
执行MSSP选点获取无标记顶点： 118.0
已标记列表 [1, 55, 103, 70, 57, 106, 89, 86, 104, 56, 67, 107, 132, 109, 46, 118]
已标记字典 OrderedDict([(0, [1]), (1, [55, 70, 57, 89, 86, 56, 67, 46]), (2, [103, 106, 104, 107, 132, 109, 118])])
截超边= [13, 102, 108]
进入S2>>>>>>>>>>>>>>>>>>>>>>>>>剩余预算4>>>>>>>>>>>>>>>>>>>>>>>>>>
执行MSSP选点获取无标记顶点： 110.0
已标记列表 [1, 55, 103, 70, 57, 106, 89, 86, 104, 56, 67, 107, 132, 109, 46, 118, 110]
已标记字典 OrderedDict([(0, [1]), (1, [55, 70, 57, 89, 86, 56, 67, 46]), (2, [103, 106, 104, 107, 132, 109, 118, 110])])
截超边= [30, 80]
进入S2>>>>>>>>>>>>>>>>>>>>>>>>>剩余预算2>>>>>>>>>>>>>>>>>>>>>>>>>>
执行MSSP选点获取无标记顶点： 112.0
已标记列表 [1, 55, 103, 70, 57, 106, 89, 86, 104, 56, 67, 107, 132, 109, 46, 118, 110, 112]
已标记字典 OrderedDict([(0, [1]), (1, [55, 70, 57, 89, 86, 56, 67, 46]), (2, [103, 106, 104, 107, 132, 109, 118, 110, 112])])
截超边= [23, 53]
剩余标记预算 0
213
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
