DeepLearning.AI TensorFlow Developer Course 1-2

Posted on | In

[Coursera] DeepLearning.AI TensorFlow Developer ( Tensorflow In Practice ) 강의 정리
Course 1 : Introduction to TensorFlow for Artificial Intelligence, Machine Learning, and Deep Learning

  • Week 1 - A New Programming Paradigm
  • Week 2 - Introduction to Computer Vision
  • Week 3 - Enhancing Vision with Convolutional Neural Networks
  • Week 4 - Using Real-world Images

Course 2 : Convolutional Neural Networks in TensorFlow
Course 3 : Natural Language Processing in TensorFlow
Course 4 : Sequences, Time Series and Prediction

Week 2 - Introduction to Computer Vision

An Introduction to computer vision

Fashion MNIST dataset

  • 70k Images
    70,000개 이미지
  • 10 Categories
    10가지 의류
  • Images are 28x28
    28 x 28 픽셀 : 각 픽셀은 0 - 255 표현
  • Can train a neural net!

Writing code to load training data

1
2
mnist = keras.datasets.fashion_mnist
(train_images, train_labels), (test_images, test_labels) = fashion_mnist.load_data()
  • 총 70,000개 이미지
  • train - 60,000개
  • test - 10,000개
레이블 클래스
0 T-shirt/top
1 Trouser
2 Pullover
3 Dress
4 Coat
5 Sandal
6 Shirt
7 Sneaker
8 Bag
9 Ankle boot

https://www.tensorflow.org/tutorials/keras/classification

Coding a Computer Vision Neural Network

CNN 모델링 준비

1
2
3
4
5
model = tf.keras.models.Sequential([
		tf.keras.layers.Flatten(input_shape=(28, 28)),
		tf.keras.layers.Dense(512, activation=tf.nn.relu),
		tf.keras.layers.Dense(10, activation=tf.nn.softmax)
])

첫번째 레이어
tf.keras.layers.Flatten(input_shape=(28, 28))
이미지가 28 x 28이므로, input shape는 28x28 flatten layer이다.

마지막 레이어
10개 뉴런이 있는 이유는 데이터 셋에 10가지 종류 의류가 있기 때문이다.

Walk through a Notebook for computer vision

이미지는 0 ~ 255 사이 값으로 표현된다.
모든 값을 0 ~ 1 변경하기 위해 255로 나눈다.

1
2
(x_train, y_train),(x_test, y_test) = mnist.load_data()
x_train, x_test = x_train / 255.0, x_test / 255.0

Using Callbacks to control training

Callback은 사용자가 지정한 시점에서 학습을 취소 할 수 있다.

1
2
3
4
5
class myCallback(tf.keras.callbacks.Callback):
	def on_epoch_end(self, epoch, logs={}):
		if(logs.get('loss')<0.4):
			print("\nLoss is low so cancelling training!")
			self.model.stop_training = True

예를 들어, 학습 중 Loss가 0.4미만이면 학습을 취소한다.

Walk through a notebook with Callbacks

전체 코드

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
import tensorflow as tf

class myCallback(tf.keras.callbacks.Callback):
	def on_epoch_end(self, epoch, logs={}):
		if(logs.get('loss')<0.4):
			print("\nLoss is low so cancelling training!")
			self.model.stop_training = True

mnist = keras.datasets.fashion_mnist
(train_images, train_labels), (test_images, test_labels) = mnist.load_data()

train_images, test_images = train_images / 255.0, test_images / 255.0

callbacks = myCallback()
model = tf.keras.models.Sequential([
		tf.keras.layers.Flatten(input_shape=(28, 28)),
		tf.keras.layers.Dense(512, activation=tf.nn.relu),
		tf.keras.layers.Dense(10, activation=tf.nn.softmax)
		])
model.compile(optimizer=tf.optimizers.Adam(),
loss='sparse_categorical_crossentropy',
metrics=['accuracy'])
model.fit(x_train, y_train, epochs=10, callbacks=[callbacks])

callbacks 사용보다는 끝까지 기다리는 것을 추천!


https://www.coursera.org/professional-certificates/tensorflow-in-practice

0%