def QueryGen(queryPoint, S, T_cols):
"""
用于用户输入的一个查询数据点，产生线性变换所需矩阵
:param queryPoint:
:return:
"""
#对用户输入的queryPoint做特殊构造以便于后文的计算
# print(queryPoint.shape)
# print( queryPoint[0])
dim = queryPoint.shape[0] #注意数据维度的获取
#对查询向量做特殊构造
newQueryPoint = np.zeros((1, dim + 2), dtype=object)
temp = queryPoint.dot(queryPoint.T)
newQueryPoint[0][0] = temp
newQueryPoint[0][1] = 1
for j in range(dim):
newQueryPoint[0][j + 2] = (-2) * queryPoint[j]
dim = newQueryPoint.shape[1] #更新dim
newQueryPoint.resize((1, dim)) # 转换数据维度
G = (np.copy(newQueryPoint))
newPs, newPm = mvhe.getinvertiblematrix(1 + T_cols)
newT = mvhe.getRandomMatrix(1, T_cols, mvhe.tBound) # 产生新随机矩阵T
newS = mvhe.getSecretKey(newT, newPs) # 新密钥
GS = G.dot(S)
M = mvhe.KeySwicthMatrix(GS, newT, newPm)
return M, newS
def QueryGen(queryDataSet, S, T_cols):
"""
用于用户输入的一个查询数据点，产生线性变换所需矩阵
:param queryPoint:
:return:
"""
# 对用户输入的queryDataSet做特殊构造以便于后文的计算
queryPoints, dim = queryDataSet.shape
queryDataReshape = np.zeros((queryPoints, dim + 2), dtype=object)
for i in range(queryPoints):
temp = queryDataSet[i].dot(queryDataSet[i].T)
queryDataReshape[i][0] = temp
queryDataReshape[i][1] = 1
for j in range(dim):
queryDataReshape[i][j + 2] = (-2) * testData[i][j]
dim = queryDataReshape.shape[1] # 更新dim
queryDataReshape.resize((queryPoints, dim)) # 转换数据维度
G = (np.copy(queryDataReshape))
newPs, newPm = mvhe.getinvertiblematrix(queryPoints + T_cols)
newT = mvhe.getRandomMatrix(queryPoints, T_cols, mvhe.tBound) # 产生新随机矩阵T
newS = mvhe.getSecretKey(newT, newPs) # 新密钥
GS = G.dot(S)
M = mvhe.KeySwicthMatrix(GS, newT, newPm)
return M, newS
def DataUpload(trainData):
trainDataNums,dim = trainData.shape
#数据构造
trainDataReshape = np.zeros((trainDataNums, dim + 2), dtype=object)
for i in range(trainDataNums):
trainDataReshape[i][0] = 1
temp = trainData[i].dot(trainData[i].T)
trainDataReshape[i][1] = temp
for j in range(dim):
trainDataReshape[i][j + 2] = trainData[i][j]
dim = trainDataReshape.shape[1] #维度更新
#加密
# 初始化参数
T_cols = 1
Ps, Pm = mvhe.getinvertiblematrix(dim + T_cols)
T = mvhe.getRandomMatrix(dim, T_cols, mvhe.tBound)
S = mvhe.getSecretKey(T, Ps)
#加密
encOftrainDataReshape = np.zeros((trainDataNums, dim + T.shape[1]), dtype=object)
for i in range(trainDataNums):
encOftrainDataReshape[i] = mvhe.encrypt(T, Pm, trainDataReshape[i])
return encOftrainDataReshape, S, T_cols
# dataset = read_points('CDR_Z.txt')
dataset = read_points('CDR_raw_2d_Z.txt')
dataset = np.array(dataset)
print("Dataset has %d items and %d dims" %
(dataset.shape[0], dataset.shape[1]))
print(dataset[0])
# 参数设置
row = dataset.shape[0]
col = dataset.shape[1]
print(row, col)
K = 1
N = col
St, Mt = mvhe.getinvertiblematrix(N + K)
T = mvhe.getRandomMatrix(N, K, mvhe.tBound)
S = mvhe.getSecretKey(T, St)
H = np.dot(S.T, S)
def main():
for i in range(2, 10):
k = i + 1
start = time.time()
vhe_kmeans(dataset, k)
end = time.time()
print((end - start) / 3600, 'hour')
t = open("vhe_time.txt", 'a+')
print((end - start) / 3600, 'hour', file=t)
t.close()