1
0

Delete article

Deleted articles cannot be recovered.

Draft of this article would be also deleted.

Are you sure you want to delete this article?

More than 5 years have passed since last update.

多項ロジスティック回帰 [TensorFlow2.0でDeep Learning 2]

Last updated at Posted at 2019-05-24

(目次はこちら)

#はじめに
多項ロジスティック回帰 [TensorFlowでDeep Learning 2]をtensorflow2.0で実現するためにはどうしたらいいのかを書く(tf.keras)。

コード

Python: 3.6.8, Tensorflow: 2.0.0a0で動作確認済み

多項ロジスティック回帰 [TensorFlowでDeep Learning 2] (mnist_softmax.py)を書き換えると、

v2/mnist_softmax.py

v2/mnist_softmax.py
from helper import *

IMAGE_SIZE = 28 * 28
CATEGORY_NUM = 10
LEARNING_RATE = 0.1
EPOCHS = 30
BATCH_SIZE = 100
LOG_DIR = 'log_softmax'
EPS = 1e-10

def loss_fn(y_true, y):
    y = tf.clip_by_value(y, EPS, 1.0)
    return -tf.reduce_sum(y_true * tf.math.log(y), axis=1)

class LR(tf.keras.layers.Layer):
    def __init__(self, units, *args, **kwargs):
        super().__init__(*args, **kwargs)
        self.units = units

    def build(self, input_shape):
        input_dim = int(input_shape[-1])
        self.W = self.add_weight(
            name='weight',
            shape=(input_dim, self.units),
            initializer=tf.keras.initializers.GlorotUniform()
        )
        self.b = self.add_weight(
            name='bias',
            shape=(self.units,),
            initializer=tf.keras.initializers.Zeros()
        )
        self.built = True

    def call(self, x):
        return tf.nn.softmax(tf.matmul(x, self.W) + self.b)

if __name__ == '__main__':
    (X_train, y_train), (X_test, y_test) = mnist_samples(flatten_image=True)

    model = tf.keras.models.Sequential()
    model.add(LR(CATEGORY_NUM, input_shape=(IMAGE_SIZE,)))
    model.compile(loss=loss_fn, optimizer=tf.keras.optimizers.SGD(LEARNING_RATE), metrics=['accuracy'])

    cb = [tf.keras.callbacks.TensorBoard(log_dir=LOG_DIR)]
    model.fit(X_train, y_train, batch_size=BATCH_SIZE, epochs=EPOCHS, callbacks=cb, validation_data=(X_test, y_test))
    print(model.evaluate(X_test, y_test))

と書ける。

ロジスティック回帰との違いは、交差エントロピーの最小化となることと、

return -tf.reduce_sum(y_true * tf.math.log(y), axis=1)

sigmoid()softmax()になるくらい。

return tf.nn.softmax(tf.matmul(x, self.W) + self.b)

前回と同様に、シンプルに書ける。

v2/mnist_softmax_simple.py

v2/mnist_softmax_simple.py
from helper import *

IMAGE_SIZE = 28 * 28
CATEGORY_NUM = 10
LEARNING_RATE = 0.1
EPOCHS = 30
BATCH_SIZE = 100
LOG_DIR = 'log_softmax'


if __name__ == '__main__':
    (X_train, y_train), (X_test, y_test) = mnist_samples(flatten_image=True)

    model = tf.keras.models.Sequential()
    model.add(tf.keras.layers.Dense(CATEGORY_NUM, input_shape=(IMAGE_SIZE,), activation='softmax'))
    model.compile(
            loss='categorical_crossentropy',
            optimizer=tf.keras.optimizers.SGD(LEARNING_RATE), metrics=['accuracy'])

    cb = [tf.keras.callbacks.TensorBoard(log_dir=LOG_DIR)]
    model.fit(X_train, y_train, batch_size=BATCH_SIZE, epochs=EPOCHS, callbacks=cb, validation_data=(X_test, y_test))
    print(model.evaluate(X_test, y_test))

めでたしめでたし
image.png

1
0
0

Register as a new user and use Qiita more conveniently

  1. You get articles that match your needs
  2. You can efficiently read back useful information
  3. You can use dark theme
What you can do with signing up
1
0

Delete article

Deleted articles cannot be recovered.

Draft of this article would be also deleted.

Are you sure you want to delete this article?