티스토리 뷰
[딥러닝 기본 20] Lab 07_1 - training / test data set, learning rate, normalization
Nero :) 2017. 6. 6. 17:24모두를 위한 머신러닝 / 딥러닝 김성훈 교수님 강의를 듣고 정리한 내용입니다.
Lab 07_1 - training / test data set, learning rate, normalization 
@ Training and Test data sets으로 분류하여 학습
- Training data sets : 학습에만 사용
- Test data sets : 모델이 한번도 본 적 없는 데이터를 가지고 공정한 학습 평가
@ TensorFlow
- Training data sets과 Test data sets 분류 예제
import os
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'
import tensorflow as tf
# 학습 데이터
x_data = [[1, 2, 1], [1, 3, 2], [1, 3, 4], [1, 5, 5], [1, 7, 5], [1, 2, 5], [1, 6, 6], [1, 7, 7]]
y_data = [[0, 0, 1], [0, 0, 1], [0, 0, 1], [0, 1, 0], [0, 1, 0], [0, 1, 0], [1, 0, 0], [1, 0, 0]]
# Test 데이터
x_test = [[2, 1, 1], [3, 1, 2], [3, 3, 4]]
y_test = [[0, 0, 1], [0, 0, 1], [0, 0, 1]]
X = tf.placeholder("float", [None, 3])
Y = tf.placeholder("float", [None, 3])
W = tf.Variable(tf.random_normal([3, 3]))
b = tf.Variable(tf.random_normal([3]))
# hypothesis
hypothesis = tf.nn.softmax(tf.matmul(X, W)+b)
cost = tf.reduce_mean(-tf.reduce_sum(Y * tf.log(hypothesis), axis=1))
optimizer = tf.train.GradientDescentOptimizer(learning_rate=0.1).minimize(cost)
# Correct prediction Test model
prediction = tf.arg_max(hypothesis, 1)
is_correct = tf.equal(prediction, tf.arg_max(Y, 1))
accuracy = tf.reduce_mean(tf.cast(is_correct, tf.float32))
# Launch graph
with tf.Session() as sess:
# Initialize TensorFlow variables
sess.run(tf.global_variables_initializer())
for step in range(201):
# 학습에는 학습 데이터 이용
cost_val, W_val, _ = sess.run([cost, W, optimizer],
feed_dict={X: x_data, Y: y_data})
print(step, cost_val, W_val)
# 200 0.582307
# [[-0.97919977 - 0.42496243 2.13029099]
# [-0.84562796 - 1.03543162 - 1.07710946]
# [0.66253269 0.44326913 - 0.41879809]]
# predict : 테스트에는 테스트 데이터 이용
print("Prediction:", sess.run(prediction, feed_dict={X: x_test}))
# Prediction: [2 2 2]
# Calculate the accuracy
print("Accuracy: ", sess.run(accuracy, feed_dict={X: x_test, Y: y_test}))
# Accuracy: 1.0
- learning rate = 1.5로 크게 준 예제
import os
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'
import tensorflow as tf
# 학습 데이터
x_data = [[1, 2, 1], [1, 3, 2], [1, 3, 4], [1, 5, 5], [1, 7, 5], [1, 2, 5], [1, 6, 6], [1, 7, 7]]
y_data = [[0, 0, 1], [0, 0, 1], [0, 0, 1], [0, 1, 0], [0, 1, 0], [0, 1, 0], [1, 0, 0], [1, 0, 0]]
# Test 데이터
x_test = [[2, 1, 1], [3, 1, 2], [3, 3, 4]]
y_test = [[0, 0, 1], [0, 0, 1], [0, 0, 1]]
X = tf.placeholder("float", [None, 3])
Y = tf.placeholder("float", [None, 3])
W = tf.Variable(tf.random_normal([3, 3]))
b = tf.Variable(tf.random_normal([3]))
# hypothesis
hypothesis = tf.nn.softmax(tf.matmul(X, W)+b)
cost = tf.reduce_mean(-tf.reduce_sum(Y * tf.log(hypothesis), axis=1))
# learning_rate = 1.5로 크게 준 경우
optimizer = tf.train.GradientDescentOptimizer(learning_rate=1.5).minimize(cost)
# Correct prediction Test model
prediction = tf.arg_max(hypothesis, 1)
is_correct = tf.equal(prediction, tf.arg_max(Y, 1))
accuracy = tf.reduce_mean(tf.cast(is_correct, tf.float32))
# Launch graph
with tf.Session() as sess:
# Initialize TensorFlow variables
sess.run(tf.global_variables_initializer())
for step in range(201):
# 학습에는 학습 데이터 이용
cost_val, W_val, _ = sess.run([cost, W, optimizer],
feed_dict={X: x_data, Y: y_data})
print(step, cost_val, W_val)
# 값이 nan으로 학습이 안됨
# 200 nan
# [[nan nan nan]
# [nan nan nan]
# [nan nan nan]]
# predict : 테스트에는 테스트 데이터 이용
print("Prediction:", sess.run(prediction, feed_dict={X: x_test}))
# Prediction: [0 0 0]
# Calculate the accuracy
print("Accuracy: ", sess.run(accuracy, feed_dict={X: x_test, Y: y_test}))
# Accuracy: 0.0
- learning rate = 1e-10 으로 작게 준 예제
import os
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'
import tensorflow as tf
# 학습 데이터
x_data = [[1, 2, 1], [1, 3, 2], [1, 3, 4], [1, 5, 5], [1, 7, 5], [1, 2, 5], [1, 6, 6], [1, 7, 7]]
y_data = [[0, 0, 1], [0, 0, 1], [0, 0, 1], [0, 1, 0], [0, 1, 0], [0, 1, 0], [1, 0, 0], [1, 0, 0]]
# Test 데이터
x_test = [[2, 1, 1], [3, 1, 2], [3, 3, 4]]
y_test = [[0, 0, 1], [0, 0, 1], [0, 0, 1]]
X = tf.placeholder("float", [None, 3])
Y = tf.placeholder("float", [None, 3])
W = tf.Variable(tf.random_normal([3, 3]))
b = tf.Variable(tf.random_normal([3]))
# hypothesis
hypothesis = tf.nn.softmax(tf.matmul(X, W)+b)
cost = tf.reduce_mean(-tf.reduce_sum(Y * tf.log(hypothesis), axis=1))
# learning_rate = 1e-10으로 작게 준 경우
optimizer = tf.train.GradientDescentOptimizer(learning_rate=0.01).minimize(cost)
# Correct prediction Test model
prediction = tf.arg_max(hypothesis, 1)
is_correct = tf.equal(prediction, tf.arg_max(Y, 1))
accuracy = tf.reduce_mean(tf.cast(is_correct, tf.float32))
# Launch graph
with tf.Session() as sess:
# Initialize TensorFlow variables
sess.run(tf.global_variables_initializer())
for step in range(201):
# 학습에는 학습 데이터 이용
cost_val, W_val, _ = sess.run([cost, W, optimizer],
feed_dict={X: x_data, Y: y_data})
print(step, cost_val, W_val)
# 학습이 거의 진전이 없음 (cost 값이 100번째나 200번째나 비슷)
# 100 3.36748
# [[-0.69622648 -1.94015765 0.01134601]
# [0.61783504 0.05631189 0.39131904]
# [-0.39081162 -0.61218679 -1.00037909]]
# 200 3.36748
# [[-0.69622648 -1.94015765 0.01134601]
# [0.61783504 0.05631189 0.39131904]
# [-0.39081162 -0.61218679 -1.00037909]]
# predict : 테스트에는 테스트 데이터 이용
print("Prediction:", sess.run(prediction, feed_dict={X: x_test}))
# Prediction: [0 0 0]
# Calculate the accuracy
print("Accuracy: ", sess.run(accuracy, feed_dict={X: x_test, Y: y_test}))
# Accuracy: 0.0
- Non-normalized inputs data의 학습 예제
import os
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'
import tensorflow as tf
import numpy as np
xy = np.array([[828.659973, 833.450012, 908100, 828.349976, 831.659973],
[823.02002, 828.070007, 1828100, 821.655029, 828.070007],
[819.929993, 824.400024, 1438100, 818.97998, 824.159973],
[816, 820.958984, 1008100, 815.48999, 819.23999],
[819.359985, 823, 1188100, 818.469971, 818.97998],
[819, 823, 1198100, 816, 820.450012],
[811.700012, 815.25, 1098100, 809.780029, 813.669983],
[809.51001, 816.659973, 1398100, 804.539978, 809.559998]])
# 0 ~ 3 : x_data
# 4 : y_data
x_data = xy[:, 0:-1]
y_data = xy[:, [-1]]
# placeholder
X = tf.placeholder(tf.float32, shape=[None, 4])
Y = tf.placeholder(tf.float32, shape=[None, 1])
W = tf.Variable(tf.random_normal([4, 1]), name='weight')
b = tf.Variable(tf.random_normal([1]), name='bias')
hypothesis = tf.matmul(X, W) + b
cost = tf.reduce_mean(tf.square(hypothesis - Y))
# Minimize
optimizer = tf.train.GradientDescentOptimizer(learning_rate=1e-5)
train = optimizer.minimize(cost)
sess = tf.Session()
sess.run(tf.global_variables_initializer())
for step in range(2001):
cost_val, hy_val, _ = sess.run([cost, hypothesis, train], feed_dict={X: x_data, Y: y_data})
print(step, "Cost: ", cost_val, "\nPrediction:\n", hy_val)
# learning_rate를 이것저것 바꿔봐도 input data의 값들간의 차이가 너무 커서 학습이 이루어지지 않음
# 2000
# Cost: nan
# Prediction:
# [[nan]
# [nan]
# [nan]
# [nan]
# [nan]
# [nan]
# [nan]
# [nan]]
- min-max scale을 이용한 normalize 학습 예제
import os
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'
import tensorflow as tf
import numpy as np
xy = np.array([[828.659973, 833.450012, 908100, 828.349976, 831.659973],
[823.02002, 828.070007, 1828100, 821.655029, 828.070007],
[819.929993, 824.400024, 1438100, 818.97998, 824.159973],
[816, 820.958984, 1008100, 815.48999, 819.23999],
[819.359985, 823, 1188100, 818.469971, 818.97998],
[819, 823, 1198100, 816, 820.450012],
[811.700012, 815.25, 1098100, 809.780029, 813.669983],
[809.51001, 816.659973, 1398100, 804.539978, 809.559998]])
# Min-Max Scale을 이용한 normalize
def MinMaxScaler(data):
numerator = data - np.min(data, 0)
denominator = np.max(data, 0) - np.min(data, 0)
# noise term prevents the zero division
return numerator / (denominator + 1e-7)
xy = MinMaxScaler(xy)
# 0 ~ 3 : x_data
# 4 : y_data
x_data = xy[:, 0:-1]
y_data = xy[:, [-1]]
# placeholder
X = tf.placeholder(tf.float32, shape=[None, 4])
Y = tf.placeholder(tf.float32, shape=[None, 1])
W = tf.Variable(tf.random_normal([4, 1]), name='weight')
b = tf.Variable(tf.random_normal([1]), name='bias')
hypothesis = tf.matmul(X, W) + b
cost = tf.reduce_mean(tf.square(hypothesis - Y))
# Minimize
optimizer = tf.train.GradientDescentOptimizer(learning_rate=0.01)
train = optimizer.minimize(cost)
sess = tf.Session()
sess.run(tf.global_variables_initializer())
for step in range(2001):
cost_val, hy_val, _ = sess.run([cost, hypothesis, train], feed_dict={X: x_data, Y: y_data})
print(step, "Cost: ", cost_val, "\nPrediction:\n", hy_val)
# MinMaxScaler를 이용하여 normalize & 학습
# 2000
# Cost: 0.0116639
# Prediction:
# [[1.09614301]
# [0.77814764]
# [0.56500143]
# [0.38963056]
# [0.47361973]
# [0.4958491]
# [0.0413903]
# [0.21422961]]
'ML (Machine Learning)' 카테고리의 다른 글
| [딥러닝 기본 22] Lab 07_2 - meet mnist dataset (정리) (0) | 2017.07.05 |
|---|---|
| [딥러닝 기본 21] Lab 07_2 - meet mnist dataset (0) | 2017.06.08 |
| [딥러닝 기본 19] Lec 07_2 - training / test 데이터 셋 (0) | 2017.06.06 |
| [딥러닝 기본 18] Lec 07_1 - learning rate, overfitting 그리고 일반화(regularization) (0) | 2017.06.06 |
| [딥러닝 기본 17] Lab 06_2 - tensorflow로 fancy softmax classification의 구현하기 (0) | 2017.06.02 |