lec5_LogisticClassification

'''

Classification

    Spam Detection : Spam(1) of Ham(0)

    Facebook feed : show(1) or hide(0)

    Credit Card Fraudulent Transaction detection : legitimate(0)/fraud(1)

    

    g(z)=1/(1+e^(-z)) <=== logistic function, sigmoid function

                            0과 1 사이

                            

    정확도가 높다.

    뉴럴 네트워크의 중요한 구성요소

    

    New cost function for logistic

        cost(H(x),y) = -log(H(x))       :y=1

                       -log(1-H(x))     :y=0

                       

        y=1                            y=0

        H(x)=1 -> cost(1)=0            H(x)=0 -> cost=0

        H(x)=0 -> cost=무한히커진다.    H(x)=1 -> cost=무한히 커진다.

        

        cost(H(x),y) = -y*log(H(x))-(1-y)*log(1-H(x))

'''

import tensorflow as tf

 

x_data = [[1,2],[2,3],[3,1],[4,3],[5,3],[6,2]]

y_data = [[0],[0],[0],[1],[1],[1]]

 

# placeholders for a tensor that will be always fed

X=tf.placeholder(tf.float32,shape=[None,2])

Y=tf.placeholder(tf.float32,shape=[None,1])

W=tf.Variable(tf.random_normal([2,1]),name='weight')

b=tf.Variable(tf.random_normal([1]),name='bias')

 

#Hypothesis using sigmoid : tf.div(1.,1.+tf.exp(tf.matmul(X,W)+b))

hypothesis = tf.sigmoid(tf.matmul(X,W)+b)

 

#cost/loss function

cost = -tf.reduce_mean(Y*tf.log(hypothesis)+(1-Y)*tf.log(1-hypothesis))

 

train = tf.train.GradientDescentOptimizer(learning_rate=0.01).minimize(cost)

 

#Accuracy computation

#True if hypothesis>0.5 else False

predicted = tf.cast(hypothesis > 0.5,dtype=tf.float32)

accuracy = tf.reduce_mean(tf.cast(tf.equal(predicted,Y),dtype=tf.float32))

 

#launch graph

with tf.Session() as sess:

    #initalize tensorflow variables

    sess.run(tf.global_variables_initializer())

    

    for step in range(10001):

        cost_val,_=sess.run([cost,train],feed_dict={X:x_data,Y:y_data})

        if step %200 == 0:

            print(step,cost_val)

            

    #Accurancy report

    h,c,a = sess.run([hypothesis,predicted,accuracy],

                    feed_dict={X:x_data,Y:y_data})

    print("\nHypothesis: ",h,"\nCorrect(Y): ",c, "\nAccuracy: ", a)