[DL] CNN - 3

CNN 기초예제_TF1.15(MNIST)

기본 CNN을 이용해 mnist 예제를 예측해본다. 여기서는 jupyter notebook이 아닌 colab을 사용했다.

Library

사용된 Library를 정의한다.

import numpy as np
import pandas as pd
import tensorflow as tf

from sklearn.preprocessing import MinMaxScaler
from sklearn.model_selection import train_test_split
from sklearn.metrics import classification_report

DataSet

DataSet를 불러오고 preprocessing 을 진행한다.

df = read_csv('/content/drive/My Drive/MachineLearning/data/mnist/train.csv')

x_data_train, x_data_test, t_data_train, t_data_test =\
train_test_split(df.drop('label', axis=1, inplace=False),
                 df['label'], test_size=0.3, random_state=0)

scaler = MaxMinScaler()
scaler.fit(x_data_train)
x_data_train_norm = scaler.transform(x_data_train)
x_data_test_norm = scaler.transform(x_data_test)

del x_data_train
del x_data_test

sess = tf.Session()

t_data_train_onehot = sess.run(tf.one_hot(t_data_train, depth=10))
t_data_test_onehot = sess.run(tf.one_hot(t_data_test, depth=10))

CNN

TF1.15를 가지고 CNN모델을 완성한다.

## placeholder
X = tf.placeholder(shape=(None,x_data_train_norm.shape[1]), dtype=tf.float32)
T = tf.placeholder(shape=(None,t_data_train_onehot.shape[1]), dtype=tf.float32)
drop_rate = tf.placeholder(dtype=tf.float32)

## 입력 데이터 형태 설정
x_img = tf.reshape(X,[-1 ,28 ,28 , 1]) # (개수, height, width, channel)
## convolution layer1
W1 = tf.variable(tf.random.normal([3,3 , 1, 32])) # (height,width,channel, 개수)

L1 = tf.conv2d(x_img, W1, strides=[1,1,1,1], pooling='SAME')
L1 = tf.nn.relu(L1)

## pooling layer1
L1 = tf.nn.max_pool(L1, ksize=[1,2,2,1] strides=[1,2,2,1], pooling='SAME')
print('L1의 shape : {}'.format(L1.shape)) # L1의 shape : (?, 14, 14, 32)

## convolution layer2
W2 = tf.variable(tf.normal.random([3,3,32,64]))

L2 = tf.nn.conv2d(L1, W2, strides=[1,1,1,1], pooling='SAME')
L2 = tf.nn.relu(L2)

## pooling layer2
L2 = tf.nn.max_pool(L2, ksize=[1,1,1,1], strides=[1,1,1,1], pooling='SAME')
print('L2의 shape : {}'.format(L2.shape)) # L2의 shape : (?, 7, 7, 64)
## FC layer로 들어가기 위한 Flatten 과정
L2 = tf.reshape(L2,[-1, 7*7*64])

## Weight3 and bias
W3 = tf.get_varaible('weight3', shape=[7*7*64, 256], initializer=tf.contrib.layers.variance_scaling_initializer())
b3 = tf.Variable(tf.random.normal([256]), name='bias3')

## layer
_layer3 = tf.nn.relu(tf.matmul(L2,w3)+b3)
layer3 = tf.nn.dropout(_layer3, rate=drop_rate)

## Weight4 and bias
W4 = tf.get_initializer('weight4', shape=[256, 10], initializer=tf.contrib.layers.variance_scaling_initializer())
b4 = tf.Variable(tf.random.normal([10]), name='bias4')

## Hypothesis
logit = tf.matmul(layer3, W4) + b4
H = tf.nn.softmax(logit)

## loss
loss = tf.reduce_mean(tf.nn.softmax_cross_validatio_with_logits_v2(logits=logit, labels=T))

## train
train = tf.train.AdamOptimizer(learning_rate=1e-3).minimize(loss)

## parameter
num_of_epoch =200
batch_size= 100

def run_train(sess,train_x,train_t):
    print('###학습이 시작되요!!###')
    sess.run(tf.global_variables_initializer())
    
    total_batch = int(train_x.shape[0]/batch_size)
    
    for step in range(num_of_epoch):
        for i in range(total_batch):
            batch_x = train_x[i*batch_size:(i+1)*batch_size]
            batch_t = train_t[i*batch_size:(i+1)*batch_size]
            _, loss_val = sess.run([train, loss], feed_dict={X:batch_x,
                                                             T:batch_t,
                                                             drop_rate:0.4})
        if step % 20 == 0:
            print('loss : {}'.format(loss_val))
    print('###학습이 종료되요!!###')
result = run_train(sess, x_data_train_norm, t_data_train_onehot)
## ###학습이 시작되요!!###
## loss : 0.4871692657470703
## loss : 0.02252289280295372
## loss : 0.052905045449733734
## loss : 0.0018550120294094086
## loss : 3.7839847209397703e-05
## loss : 0.000732154177967459
## loss : 1.346974613625207e-06
## loss : 5.364403321550526e-08
## loss : 1.19209286886246e-09
## loss : 6.675698216440651e-08
## ###학습이 종료되요!!###

## Accuracy
predict = tf.argmax(H,1)

target_names ={'num 0','num 1','num 2','num 3','num 4','num 5','num 6','num 7','num 8','num 9'}
print(classification_report(t_data_test,
                            sess.run(predict,
                                     feed_dict={X:x_data_test_norm, 
                                                drop_rate:0 }),
                            target_names=target_names))  

##               precision    recall  f1-score   support
## 
##        num 0       0.99      0.99      0.99      1242
##        num 9       0.99      0.99      0.99      1429
##        num 8       0.98      0.99      0.99      1276
##        num 1       0.99      0.99      0.99      1298
##        num 3       0.99      0.98      0.98      1236
##        num 4       0.99      0.99      0.99      1119
##        num 5       0.99      1.00      0.99      1243
##        num 6       0.99      0.99      0.99      1334
##        num 7       0.99      0.98      0.99      1204
##        num 2       0.98      0.98      0.98      1219
## 
##     accuracy                           0.99     12600
##    macro avg       0.99      0.99      0.99     12600
## weighted avg       0.99      0.99      0.99     12600