'''
auther:Deniu He
date:2020-09-30
'''
from copy import deepcopy
import numpy as np
import matplotlib.pyplot as plt
from sklearn import datasets
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score
from sklearn.linear_model import LogisticRegression
from scipy.spatial.distance import pdist,squareform
import cvxpy as cvx
class LRLR():
def __init__(self, X_labeled=None, y_labeled=None ,adjMatrix=None, X_pool=None):
self.X = None
self.y = None
self.d = None
###-------------------------------
self.X_pool = None
self.adjMatrix = None
self.N = None
self.lapMatrix = None
###------------------------------
self.lambd1 = None
self.lambd2 = None
self.w = None
self.P = None
###------------------------------
self.loglikehood = None
self.L2regu = None
self.Lapregu = None
self.obj = None
self.prob = None
def fit(self,X_labeled,y_labeled,adjMatrix,X_pool):
self.X = X_labeled
self.y = np.vstack(y_labeled)
self.d = X_labeled.shape[1]
###-------------------------------
self.X_pool = X_pool
self.adjMatrix = adjMatrix
self.N = adjMatrix.shape[0]
self.lapMatrix = np.diag(np.sum(adjMatrix,0)) - adjMatrix
###------------------------------
self.lambd1 = 0.1
self.lambd2 = 0.1
self.w = cvx.Variable((self.d, 1))
self.P = self.X_pool @ self.w
###------------------------------
self.loglikehood = cvx.sum(cvx.multiply(self.y, self.X @ self.w) - cvx.logistic(self.X @ self.w))
self.L2regu = self.lambd1 * cvx.pnorm(self.w, p=2)**2
self.Lapregu = self.lambd2 * cvx.quad_form(self.P,self.lapMatrix)
self.obj = cvx.Maximize(self.loglikehood - self.Lapregu - self.L2regu)
self.prob = cvx.Problem(self.obj).solve()
def pred_prob(self,X_test):
prob = 1 / (1 + cvx.exp(-(X_test @ self.w)).value)
return prob
def pred(self, X_test):
result = []
for ele in 1 / (1 + cvx.exp(-(X_test @ self.w)).value):
if ele >= 0.5:
result.append(1)
else:
result.append(0)
return np.array(result)
if __name__ == '__main__':
X, y = datasets.make_blobs(n_samples=500, n_features=2, centers=2, cluster_std=[3, 3], random_state=123)
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.992,random_state=1)
print("测试数据个数=",len(y_test))
plt.scatter(X[:,0],X[:,1],c=y)
plt.scatter(X_train[:,0],X_train[:,1],c='r',marker="*",s=200,label="Labeled instance")
plt.legend()
plt.show()
distMatrix = squareform(pdist(X,metric='euclidean'))
def adjacency_matrix(X):
n = X.shape[0]
nn_matrix = np.zeros((n,5))
for i in range(n):
ord_idx = np.argsort(distMatrix[i])
neibor = []
for j in range(5+1):
if i != ord_idx[j]:
neibor.append(ord_idx[j])
neibor = np.array(neibor)
nn_matrix[i] = neibor
adj_matrix = np.zeros((n,n))
for i in range(n-1):
for j in range(i,n):
if i in nn_matrix[j] or j in nn_matrix[i]:
adj_matrix[i,j] = 1
adj_matrix[j,i] = 1
return adj_matrix
adjMatrix = adjacency_matrix(X=X)
model = LRLR()
model.fit(X_labeled=X_train,y_labeled=y_train,adjMatrix=adjMatrix,X_pool=X)
y_pred = model.pred(X_test=X_test)
acc = accuracy_score(y_true=y_test,y_pred=y_pred)
print("精度：",acc)
model2 = LogisticRegression()
model2.fit(X=X_train,y=y_train)
y_pred2 = model2.predict(X=X_test)
acc2 = accuracy_score(y_true=y_test,y_pred=y_pred2)
print("调包精度：",acc2)
