大家好,又见面了,我是你们的朋友全栈君。
在我们使用k-NN模型时,需要计算测试集中每一点到训练集中每一点的欧氏距离,即需要求得两矩阵之间的欧氏距离。在实现k-NN算法时通常有三种方案,分别是使用两层循环,使用一层循环和不使用循环。
使用两层循环
分别对训练集和测试集中的数据进行循环遍历,计算每两个点之间的欧式距离,然后赋值给dist矩阵。此算法没有经过任何优化。
num_test = X.shape[0]
num_train = self.X_train.shape[0]
dists = np.zeros((num_test, num_train))
for i in xrange(num_test):
for j in xrange(num_train):
##################################################################### # TODO: # # Compute the l2 distance between the ith test point and the jth # # training point, and store the result in dists[i, j]. You should # # not use a loop over dimension. # #####################################################################
# pass
dists[i][j] = np.sqrt(np.sum(np.square(X[i] - self.X_train[j])))
##################################################################### # END OF YOUR CODE # #####################################################################
return dists
使用一层循环
使用矩阵表示训练集的数据,计算测试集中每一点到训练集矩阵的距离,可以对算法优化为只使用一层循环。
def compute_distances_one_loop(self, X):
""" Compute the distance between each test point in X and each training point in self.X_train using a single loop over the test data. Input / Output: Same as compute_distances_two_loops """
num_test = X.shape[0]
num_train = self.X_train.shape[0]
dists = np.zeros((num_test, num_train))
for i in xrange(num_test):
####################################################################### # TODO: # # Compute the l2 distance between the ith test point and all training # # points, and store the result in dists[i, :]. # #######################################################################
# pass
dists[i] = np.sqrt(np.sum(np.square(self.X_train - X[i]), axis = 1))
####################################################################### # END OF YOUR CODE # #######################################################################
return dists
不使用循环
运算效率最高的算法是将训练集和测试集都使用矩阵表示,然后使用矩阵运算的方法替代之前的循环操作。但此操作需要我们对矩阵的运算规则非常熟悉。接下来着重记录如何计算两个矩阵之间的欧式距离。
记录测试集矩阵P的大小为M*D,训练集矩阵C的大小为N*D(测试集中共有M个点,每个点为D维特征向量。训练集中共有N个点,每个点为D维特征向量)
记 Pi P i 是P的第i行,记 Cj C j 是C的第j行
Pi=[Pi1Pi2⋯PiD] P i = [ P i 1 P i 2 ⋯ P i D ] Cj=[Cj1Cj2⋯CjD] C j = [ C j 1 C j 2 ⋯ C j D ]
首先计算 Pi P i 和 Cj C j 之间的距离dist(i,j)
d(Pi,Cj)=(Pi1−Cj1)2+(Pi2−Cj2)2+⋯+(PiD−CjD)2−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−√=(P2i1+P2i2+⋯+P2iD)+(C2j1+C2j2+⋯+C2jD)−2×(Pi1Cj1+Pi2Cj2+⋯+PiDCiD)−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−√=∥Pi∥2+∥Cj∥2−2×PiCTj−−−−−−−−−−−−−−−−−−−−−√ d ( P i , C j ) = ( P i 1 − C j 1 ) 2 + ( P i 2 − C j 2 ) 2 + ⋯ + ( P i D − C j D ) 2 = ( P i 1 2 + P i 2 2 + ⋯ + P i D 2 ) + ( C j 1 2 + C j 2 2 + ⋯ + C j D 2 ) − 2 × ( P i 1 C j 1 + P i 2 C j 2 + ⋯ + P i D C i D ) = ‖ P i ‖ 2 + ‖ C j ‖ 2 − 2 × P i C j T
我们可以推广到距离矩阵的第i行的计算公式
dist[i]=(∥Pi∥2∥Pi∥2⋯∥Pi∥2)+(∥C1∥2∥C2∥2⋯∥CN∥2)−2×Pi(CT1CT2⋯CTN)−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−√=(∥Pi∥2∥Pi∥2⋯∥Pi∥2)+(∥C1∥2∥C2∥2⋯∥CN∥2)−2×PiCT−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−√ d i s t [ i ] = ( ‖ P i ‖ 2 ‖ P i ‖ 2 ⋯ ‖ P i ‖ 2 ) + ( ‖ C 1 ‖ 2 ‖ C 2 ‖ 2 ⋯ ‖ C N ‖ 2 ) − 2 × P i ( C 1 T C 2 T ⋯ C N T ) = ( ‖ P i ‖ 2 ‖ P i ‖ 2 ⋯ ‖ P i ‖ 2 ) + ( ‖ C 1 ‖ 2 ‖ C 2 ‖ 2 ⋯ ‖ C N ‖ 2 ) − 2 × P i C T
继续将公式推广为整个距离矩阵
dist=⎛⎝⎜⎜⎜⎜⎜⎜∥P1∥2∥P2∥2⋮∥PM∥2∥P1∥2∥P2∥2⋮∥PM∥2⋯⋯⋱⋯∥P1∥2∥P2∥2⋮∥PM∥2⎞⎠⎟⎟⎟⎟⎟⎟+⎛⎝⎜⎜⎜⎜⎜⎜∥C1∥2∥C1∥2⋮∥C1∥2∥C2∥2∥C2∥2⋮∥C2∥2⋯⋯⋱⋯∥CN∥2∥CN∥2⋮∥CN∥2⎞⎠⎟⎟⎟⎟⎟⎟−2×PCT−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−⎷ d i s t = ( ‖ P 1 ‖ 2 ‖ P 1 ‖ 2 ⋯ ‖ P 1 ‖ 2 ‖ P 2 ‖ 2 ‖ P 2 ‖ 2 ⋯ ‖ P 2 ‖ 2 ⋮ ⋮ ⋱ ⋮ ‖ P M ‖ 2 ‖ P M ‖ 2 ⋯ ‖ P M ‖ 2 ) + ( ‖ C 1 ‖ 2 ‖ C 2 ‖ 2 ⋯ ‖ C N ‖ 2 ‖ C 1 ‖ 2 ‖ C 2 ‖ 2 ⋯ ‖ C N ‖ 2 ⋮ ⋮ ⋱ ⋮ ‖ C 1 ‖ 2 ‖ C 2 ‖ 2 ⋯ ‖ C N ‖ 2 ) − 2 × P C T
表示为python代码:
def compute_distances_no_loops(self, X):
""" Compute the distance between each test point in X and each training point in self.X_train using no explicit loops. Input / Output: Same as compute_distances_two_loops """
num_test = X.shape[0]
num_train = self.X_train.shape[0]
dists = np.zeros((num_test, num_train))
#########################################################################
# TODO: #
# Compute the l2 distance between all test points and all training #
# points without using any explicit loops, and store the result in #
# dists. #
# #
# You should implement this function using only basic array operations; #
# in particular you should not use functions from scipy. #
# #
# HINT: Try to formulate the l2 distance using matrix multiplication #
# and two broadcast sums. #
#########################################################################
# pass
dists = np.sqrt(-2*np.dot(X, self.X_train.T) + np.sum(np.square(self.X_train), axis = 1) + np.transpose([np.sum(np.square(X), axis = 1)]))
#########################################################################
# END OF YOUR CODE #
#########################################################################
return dists
发布者:全栈程序员-用户IM,转载请注明出处:https://javaforall.cn/150938.html原文链接:https://javaforall.cn
【正版授权,激活自己账号】: Jetbrains全家桶Ide使用,1年售后保障,每天仅需1毛
【官方授权 正版激活】: 官方授权 正版激活 支持Jetbrains家族下所有IDE 使用个人JB账号...