"""
* Created with PyCharm
* 作者: 阿光
* 日期: 2022/1/2
* 时间: 18:32
* 描述:
"""
import tensorflow as tf
import tensorflow.keras.datasets.mnist
from keras import Input, Model
from keras.layers import Dense
from tensorflow import keras
(train_images, train_labels), (val_images, val_labels) = tensorflow.keras.datasets.mnist.load_data()
train_images, val_images = train_images / 255.0, val_images / 255.0
train_images = train_images.reshape(60000, 784)
val_images = val_images.reshape(10000, 784)
class CategoricalTruePositives(keras.metrics.Metric):
def __init__(self, name="categorical_true_positives"):
super(CategoricalTruePositives, self).__init__(name=name)
self.true_positives = self.add_weight(name='ctp', initializer='zeros')
def update_state(self, y_true, y_pred, sample_weight=None):
y_pred = tf.reshape(tf.argmax(y_pred, axis=1), shape=(-1, 1))
values = tf.cast(y_true, 'int32') == tf.cast(y_pred, 'int32')
values = tf.cast(values, 'float32')
self.true_positives.assign_add(tf.reduce_sum(values))
def result(self):
return self.true_positives
def reset_states(self):
self.true_positives.assign(0.0)
def get_model():
inputs = Input(shape=(784,))
outputs = Dense(10, activation='softmax')(inputs)
model = Model(inputs, outputs)
model.compile(
optimizer=keras.optimizers.RMSprop(learning_rate=1e-3),
loss=keras.losses.SparseCategoricalCrossentropy(),
metrics=[CategoricalTruePositives()]
)
return model
model = get_model()
model.fit(
train_images,
train_labels,
epochs=5,
batch_size=32,
validation_data=(val_images, val_labels)
)
