Mercurial > hg > tvii

comparison tvii/logistic_regression.py @ 55:0908b6cd3217

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
[regression] add better cost function for sigmoids
author Jeff Hammel <k0scist@gmail.com>
date Sun, 24 Sep 2017 15:30:15 -0700
parents 0807ac8992ba
children
comparison
equal deleted inserted replaced
54:0807ac8992ba 55:0908b6cd3217
96 return (-1./m)*cost 96 return (-1./m)*cost
97 97
98 def compute_costs(Yhat, Y): 98 def compute_costs(Yhat, Y):
99 """ 99 """
100 Computes the cross-entropy cost given: 100 Computes the cross-entropy cost given:
101 """ 101
102 raise NotImplementedError('TODO') 102 J = -(1/m)*sum_{i=0..m} (y(i)log(yhat(i)) + (1 - y(i))log(1 - yhat(i)))
103
104 Yhat -- The sigmoid output of the network
105 Y -- "true" label vector
106 """
107
108 # compute the cross-entropy cost
109 logprops = np.multiply(np.log(Yhat, Y)) + np.multiply(np.log(1-Yhat), (1-Y))
110 cost = - np.sum(logprobs)/m
111
112 cost = np.squeeze(cost) # make sure cost is the dimension we expect
113 assert (isinstance(cost, float))
114 return cost
103 115
104 116
105 def optimize(w, b, X, Y, num_iterations, learning_rate, print_cost = False): 117 def optimize(w, b, X, Y, num_iterations, learning_rate, print_cost = False):
106 """ 118 """
107 This function optimizes w and b by running a gradient descent algorithm 119 This function optimizes w and b by running a gradient descent algorithm