本文在调参记录6的基础上，继续调整超参数，测试Adaptively Parametric ReLU（APReLU）激活函数在Cifar10图像集上的效果。
深度残差网络 自适应参数化ReLU激活函数（调参记录6）
https://blog.csdn.net/dangqing1988/article/details/105628681
自适应参数化ReLU激活函数的基本原理见下图：
自适应参数化ReLU激活函数在Keras里，Batch Normalization的momentum默认为0.99，现在设置为0.9，这是因为momentum=0.9似乎更常见。原先Batch Normalization默认没有正则化，现在加上L2正则化，来减小过拟合。
Keras程序如下：
代码语言：javascript复制#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Created on Tue Apr 14 04:17:45 2020
Implemented using TensorFlow 1.0.1 and Keras 2.2.1
Minghang Zhao, Shisheng Zhong, Xuyun Fu, Baoping Tang, Shaojiang Dong, Michael Pecht,
Deep Residual Networks with Adaptively Parametric Rectifier Linear Units for Fault Diagnosis,
IEEE Transactions on Industrial Electronics, 2020, DOI: 10.1109/TIE.2020.2972458
@author: Minghang Zhao
"""
from __future__ import print_function
import keras
import numpy as np
from keras.datasets import cifar10
from keras.layers import Dense, Conv2D, BatchNormalization, Activation, Minimum
from keras.layers import AveragePooling2D, Input, GlobalAveragePooling2D, Concatenate, Reshape
from keras.regularizers import l2
from keras import backend as K
from keras.models import Model
from keras import optimizers
from keras.preprocessing.image import ImageDataGenerator
from keras.callbacks import LearningRateScheduler
K.set_learning_phase(1)
# The data, split between train and test sets
(x_train, y_train), (x_test, y_test) = cifar10.load_data()
# Noised data
x_train = x_train.astype('float32') / 255.
x_test = x_test.astype('float32') / 255.
x_test = x_test-np.mean(x_train)
x_train = x_train-np.mean(x_train)
print('x_train shape:', x_train.shape)
print(x_train.shape[0], 'train samples')
print(x_test.shape[0], 'test samples')
# convert class vectors to binary class matrices
y_train = keras.utils.to_categorical(y_train, 10)
y_test = keras.utils.to_categorical(y_test, 10)
# Schedule the learning rate, multiply 0.1 every 300 epoches
def scheduler(epoch):
if epoch % 300 == 0 and epoch != 0:
lr = K.get_value(model.optimizer.lr)
K.set_value(model.optimizer.lr, lr * 0.1)
print("lr changed to {}".format(lr * 0.1))
return K.get_value(model.optimizer.lr)
# An adaptively parametric rectifier linear unit (APReLU)
def aprelu(inputs):
# get the number of channels
channels = inputs.get_shape().as_list()[-1]
# get a zero feature map
zeros_input = keras.layers.subtract([inputs, inputs])
# get a feature map with only positive features
pos_input = Activation('relu')(inputs)
# get a feature map with only negative features
neg_input = Minimum()([inputs,zeros_input])
# define a network to obtain the scaling coefficients
scales_p = GlobalAveragePooling2D()(pos_input)
scales_n = GlobalAveragePooling2D()(neg_input)
scales = Concatenate()([scales_n, scales_p])
scales = Dense(channels, activation='linear', kernel_initializer='he_normal', kernel_regularizer=l2(1e-4))(scales)
scales = BatchNormalization(momentum=0.9, gamma_regularizer=l2(1e-4))(scales)
scales = Activation('relu')(scales)
scales = Dense(channels, activation='linear', kernel_initializer='he_normal', kernel_regularizer=l2(1e-4))(scales)
scales = BatchNormalization(momentum=0.9, gamma_regularizer=l2(1e-4))(scales)
scales = Activation('sigmoid')(scales)
scales = Reshape((1,1,channels))(scales)
# apply a paramtetric relu
neg_part = keras.layers.multiply([scales, neg_input])
return keras.layers.add([pos_input, neg_part])
# Residual Block
def residual_block(incoming, nb_blocks, out_channels, downsample=False,
downsample_strides=2):
residual = incoming
in_channels = incoming.get_shape().as_list()[-1]
for i in range(nb_blocks):
identity = residual
if not downsample:
downsample_strides = 1
residual = BatchNormalization(momentum=0.9, gamma_regularizer=l2(1e-4))(residual)
residual = aprelu(residual)
residual = Conv2D(out_channels, 3, strides=(downsample_strides, downsample_strides),
padding='same', kernel_initializer='he_normal',
kernel_regularizer=l2(1e-4))(residual)
residual = BatchNormalization(momentum=0.9, gamma_regularizer=l2(1e-4))(residual)
residual = aprelu(residual)
residual = Conv2D(out_channels, 3, padding='same', kernel_initializer='he_normal',
kernel_regularizer=l2(1e-4))(residual)
# Downsampling
if downsample_strides > 1:
identity = AveragePooling2D(pool_size=(1,1), strides=(2,2))(identity)
# Zero_padding to match channels
if in_channels != out_channels:
zeros_identity = keras.layers.subtract([identity, identity])
identity = keras.layers.concatenate([identity, zeros_identity])
in_channels = out_channels
residual = keras.layers.add([residual, identity])
return residual
# define and train a model
inputs = Input(shape=(32, 32, 3))
net = Conv2D(16, 3, padding='same', kernel_initializer='he_normal', kernel_regularizer=l2(1e-4))(inputs)
net = residual_block(net, 9, 16, downsample=False)
net = residual_block(net, 1, 32, downsample=True)
net = residual_block(net, 8, 32, downsample=False)
net = residual_block(net, 1, 64, downsample=True)
net = residual_block(net, 8, 64, downsample=False)
net = BatchNormalization(momentum=0.9, gamma_regularizer=l2(1e-4))(net)
net = Activation('relu')(net)
net = GlobalAveragePooling2D()(net)
outputs = Dense(10, activation='softmax', kernel_initializer='he_normal', kernel_regularizer=l2(1e-4))(net)
model = Model(inputs=inputs, outputs=outputs)
sgd = optimizers.SGD(lr=0.1, decay=0., momentum=0.9, nesterov=True)
model.compile(loss='categorical_crossentropy', optimizer=sgd, metrics=['accuracy'])
# data augmentation
datagen = ImageDataGenerator(
# randomly rotate images in the range (deg 0 to 180)
rotation_range=30,
# randomly flip images
horizontal_flip=True,
# randomly shift images horizontally
width_shift_range=0.125,
# randomly shift images vertically
height_shift_range=0.125)
reduce_lr = LearningRateScheduler(scheduler)
# fit the model on the batches generated by datagen.flow().
model.fit_generator(datagen.flow(x_train, y_train, batch_size=100),
validation_data=(x_test, y_test), epochs=1000,
verbose=1, callbacks=[reduce_lr], workers=4)
# get results
K.set_learning_phase(0)
DRSN_train_score = model.evaluate(x_train, y_train, batch_size=100, verbose=0)
print('Train loss:', DRSN_train_score[0])
print('Train accuracy:', DRSN_train_score[1])
DRSN_test_score = model.evaluate(x_test, y_test, batch_size=100, verbose=0)
print('Test loss:', DRSN_test_score[0])
print('Test accuracy:', DRSN_test_score[1])实验结果如下：
代码语言：javascript复制Epoch 755/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1093 - acc: 0.9992 - val_loss: 0.3584 - val_acc: 0.9337
Epoch 756/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1093 - acc: 0.9990 - val_loss: 0.3583 - val_acc: 0.9346
Epoch 757/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1095 - acc: 0.9991 - val_loss: 0.3560 - val_acc: 0.9346
Epoch 758/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1090 - acc: 0.9991 - val_loss: 0.3587 - val_acc: 0.9346
Epoch 759/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1092 - acc: 0.9989 - val_loss: 0.3594 - val_acc: 0.9346
Epoch 760/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1086 - acc: 0.9992 - val_loss: 0.3560 - val_acc: 0.9345
Epoch 761/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1081 - acc: 0.9993 - val_loss: 0.3573 - val_acc: 0.9346
Epoch 762/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.1083 - acc: 0.9992 - val_loss: 0.3598 - val_acc: 0.9343
Epoch 763/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1080 - acc: 0.9991 - val_loss: 0.3590 - val_acc: 0.9341
Epoch 764/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1076 - acc: 0.9993 - val_loss: 0.3567 - val_acc: 0.9336
Epoch 765/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1077 - acc: 0.9991 - val_loss: 0.3556 - val_acc: 0.9375
Epoch 766/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1072 - acc: 0.9993 - val_loss: 0.3562 - val_acc: 0.9349
Epoch 767/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1075 - acc: 0.9992 - val_loss: 0.3538 - val_acc: 0.9364
Epoch 768/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1071 - acc: 0.9991 - val_loss: 0.3607 - val_acc: 0.9347
Epoch 769/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1067 - acc: 0.9994 - val_loss: 0.3626 - val_acc: 0.9348
Epoch 770/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1070 - acc: 0.9991 - val_loss: 0.3595 - val_acc: 0.9364
Epoch 771/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1067 - acc: 0.9991 - val_loss: 0.3578 - val_acc: 0.9353
Epoch 772/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1066 - acc: 0.9991 - val_loss: 0.3561 - val_acc: 0.9357
Epoch 773/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1062 - acc: 0.9992 - val_loss: 0.3555 - val_acc: 0.9357
Epoch 774/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1062 - acc: 0.9992 - val_loss: 0.3546 - val_acc: 0.9367
Epoch 775/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1059 - acc: 0.9992 - val_loss: 0.3570 - val_acc: 0.9367
Epoch 776/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1061 - acc: 0.9990 - val_loss: 0.3570 - val_acc: 0.9355
Epoch 777/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.1065 - acc: 0.9988 - val_loss: 0.3569 - val_acc: 0.9361
Epoch 778/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.1055 - acc: 0.9991 - val_loss: 0.3592 - val_acc: 0.9347
Epoch 779/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.1053 - acc: 0.9991 - val_loss: 0.3578 - val_acc: 0.9345
Epoch 780/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1057 - acc: 0.9990 - val_loss: 0.3550 - val_acc: 0.9361
Epoch 781/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1054 - acc: 0.9988 - val_loss: 0.3598 - val_acc: 0.9359
Epoch 782/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1053 - acc: 0.9988 - val_loss: 0.3548 - val_acc: 0.9349
Epoch 783/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.1047 - acc: 0.9992 - val_loss: 0.3541 - val_acc: 0.9366
Epoch 784/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1048 - acc: 0.9990 - val_loss: 0.3540 - val_acc: 0.9346
Epoch 785/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1046 - acc: 0.9991 - val_loss: 0.3534 - val_acc: 0.9350
Epoch 786/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1041 - acc: 0.9992 - val_loss: 0.3559 - val_acc: 0.9349
Epoch 787/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1042 - acc: 0.9992 - val_loss: 0.3547 - val_acc: 0.9336
Epoch 788/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1039 - acc: 0.9992 - val_loss: 0.3523 - val_acc: 0.9347
Epoch 789/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1037 - acc: 0.9991 - val_loss: 0.3487 - val_acc: 0.9375
Epoch 790/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1034 - acc: 0.9992 - val_loss: 0.3481 - val_acc: 0.9365
Epoch 791/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1034 - acc: 0.9992 - val_loss: 0.3514 - val_acc: 0.9370
Epoch 792/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1034 - acc: 0.9991 - val_loss: 0.3507 - val_acc: 0.9363
Epoch 793/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1029 - acc: 0.9992 - val_loss: 0.3531 - val_acc: 0.9358
Epoch 794/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.1032 - acc: 0.9990 - val_loss: 0.3563 - val_acc: 0.9351
Epoch 795/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.1026 - acc: 0.9992 - val_loss: 0.3529 - val_acc: 0.9362
Epoch 796/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1024 - acc: 0.9992 - val_loss: 0.3511 - val_acc: 0.9360
Epoch 797/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1023 - acc: 0.9990 - val_loss: 0.3520 - val_acc: 0.9358
Epoch 798/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1023 - acc: 0.9990 - val_loss: 0.3524 - val_acc: 0.9354
Epoch 799/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1022 - acc: 0.9991 - val_loss: 0.3547 - val_acc: 0.9349
Epoch 800/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1020 - acc: 0.9991 - val_loss: 0.3548 - val_acc: 0.9356
Epoch 801/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.1016 - acc: 0.9993 - val_loss: 0.3524 - val_acc: 0.9356
Epoch 802/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1016 - acc: 0.9992 - val_loss: 0.3516 - val_acc: 0.9360
Epoch 803/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1015 - acc: 0.9991 - val_loss: 0.3497 - val_acc: 0.9353
Epoch 804/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.1012 - acc: 0.9992 - val_loss: 0.3520 - val_acc: 0.9355
Epoch 805/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1014 - acc: 0.9991 - val_loss: 0.3539 - val_acc: 0.9354
Epoch 806/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1010 - acc: 0.9990 - val_loss: 0.3580 - val_acc: 0.9352
Epoch 807/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1011 - acc: 0.9990 - val_loss: 0.3513 - val_acc: 0.9349
Epoch 808/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1006 - acc: 0.9992 - val_loss: 0.3521 - val_acc: 0.9367
Epoch 809/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.1005 - acc: 0.9991 - val_loss: 0.3495 - val_acc: 0.9368
Epoch 810/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.1008 - acc: 0.9988 - val_loss: 0.3529 - val_acc: 0.9350
Epoch 811/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.1001 - acc: 0.9992 - val_loss: 0.3569 - val_acc: 0.9358
Epoch 812/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0998 - acc: 0.9991 - val_loss: 0.3532 - val_acc: 0.9355
Epoch 813/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.0996 - acc: 0.9992 - val_loss: 0.3559 - val_acc: 0.9347
Epoch 814/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0997 - acc: 0.9992 - val_loss: 0.3532 - val_acc: 0.9345
Epoch 815/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0996 - acc: 0.9991 - val_loss: 0.3544 - val_acc: 0.9340
Epoch 816/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0991 - acc: 0.9991 - val_loss: 0.3529 - val_acc: 0.9358
Epoch 817/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0984 - acc: 0.9995 - val_loss: 0.3508 - val_acc: 0.9365
Epoch 818/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0994 - acc: 0.9989 - val_loss: 0.3533 - val_acc: 0.9362
Epoch 819/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.0987 - acc: 0.9993 - val_loss: 0.3519 - val_acc: 0.9351
Epoch 820/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0988 - acc: 0.9991 - val_loss: 0.3528 - val_acc: 0.9352
Epoch 821/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0983 - acc: 0.9992 - val_loss: 0.3479 - val_acc: 0.9354
Epoch 822/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0984 - acc: 0.9991 - val_loss: 0.3485 - val_acc: 0.9367
Epoch 823/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0985 - acc: 0.9990 - val_loss: 0.3530 - val_acc: 0.9358
Epoch 824/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0981 - acc: 0.9992 - val_loss: 0.3464 - val_acc: 0.9377
Epoch 825/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0978 - acc: 0.9993 - val_loss: 0.3477 - val_acc: 0.9358
Epoch 826/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0973 - acc: 0.9992 - val_loss: 0.3468 - val_acc: 0.9364
Epoch 827/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0979 - acc: 0.9991 - val_loss: 0.3502 - val_acc: 0.9358
Epoch 828/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0974 - acc: 0.9991 - val_loss: 0.3470 - val_acc: 0.9356
Epoch 829/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0969 - acc: 0.9994 - val_loss: 0.3459 - val_acc: 0.9351
Epoch 830/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0970 - acc: 0.9990 - val_loss: 0.3528 - val_acc: 0.9347
Epoch 831/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0969 - acc: 0.9992 - val_loss: 0.3484 - val_acc: 0.9360
Epoch 832/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.0970 - acc: 0.9992 - val_loss: 0.3542 - val_acc: 0.9353
Epoch 833/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0969 - acc: 0.9990 - val_loss: 0.3496 - val_acc: 0.9345
Epoch 834/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0970 - acc: 0.9990 - val_loss: 0.3460 - val_acc: 0.9372
Epoch 835/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.0960 - acc: 0.9993 - val_loss: 0.3514 - val_acc: 0.9349
Epoch 836/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0962 - acc: 0.9994 - val_loss: 0.3420 - val_acc: 0.9376
Epoch 837/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0960 - acc: 0.9992 - val_loss: 0.3441 - val_acc: 0.9358
Epoch 838/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0957 - acc: 0.9993 - val_loss: 0.3474 - val_acc: 0.9368
Epoch 839/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0955 - acc: 0.9993 - val_loss: 0.3447 - val_acc: 0.9355
Epoch 840/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.0951 - acc: 0.9995 - val_loss: 0.3508 - val_acc: 0.9355
Epoch 841/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0951 - acc: 0.9993 - val_loss: 0.3488 - val_acc: 0.9366
Epoch 842/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0952 - acc: 0.9992 - val_loss: 0.3500 - val_acc: 0.9368
Epoch 843/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.0952 - acc: 0.9991 - val_loss: 0.3464 - val_acc: 0.9359
Epoch 844/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.0947 - acc: 0.9993 - val_loss: 0.3470 - val_acc: 0.9365
Epoch 845/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.0947 - acc: 0.9993 - val_loss: 0.3478 - val_acc: 0.9353
Epoch 846/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0952 - acc: 0.9990 - val_loss: 0.3501 - val_acc: 0.9355
Epoch 847/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0944 - acc: 0.9993 - val_loss: 0.3463 - val_acc: 0.9354
Epoch 848/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0944 - acc: 0.9993 - val_loss: 0.3504 - val_acc: 0.9351
Epoch 849/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0941 - acc: 0.9993 - val_loss: 0.3468 - val_acc: 0.9373
Epoch 850/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.0947 - acc: 0.9988 - val_loss: 0.3432 - val_acc: 0.9378
Epoch 851/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.0943 - acc: 0.9989 - val_loss: 0.3456 - val_acc: 0.9369
Epoch 852/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.0943 - acc: 0.9990 - val_loss: 0.3471 - val_acc: 0.9365
Epoch 853/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0940 - acc: 0.9990 - val_loss: 0.3506 - val_acc: 0.9356
Epoch 854/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0936 - acc: 0.9992 - val_loss: 0.3498 - val_acc: 0.9358
Epoch 855/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0934 - acc: 0.9992 - val_loss: 0.3469 - val_acc: 0.9361
Epoch 856/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0931 - acc: 0.9993 - val_loss: 0.3483 - val_acc: 0.9361
Epoch 857/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0930 - acc: 0.9993 - val_loss: 0.3440 - val_acc: 0.9350
Epoch 858/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.0930 - acc: 0.9991 - val_loss: 0.3445 - val_acc: 0.9365
Epoch 859/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0928 - acc: 0.9992 - val_loss: 0.3465 - val_acc: 0.9366
Epoch 860/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0928 - acc: 0.9990 - val_loss: 0.3527 - val_acc: 0.9345
Epoch 861/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.0924 - acc: 0.9992 - val_loss: 0.3465 - val_acc: 0.9369
Epoch 862/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0923 - acc: 0.9992 - val_loss: 0.3445 - val_acc: 0.9366
Epoch 863/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0923 - acc: 0.9992 - val_loss: 0.3476 - val_acc: 0.9362
Epoch 864/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0920 - acc: 0.9993 - val_loss: 0.3454 - val_acc: 0.9369
Epoch 865/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0922 - acc: 0.9990 - val_loss: 0.3486 - val_acc: 0.9337
Epoch 866/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0914 - acc: 0.9994 - val_loss: 0.3489 - val_acc: 0.9355
Epoch 867/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.0918 - acc: 0.9991 - val_loss: 0.3467 - val_acc: 0.9359
Epoch 868/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0918 - acc: 0.9992 - val_loss: 0.3486 - val_acc: 0.9348
Epoch 869/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0913 - acc: 0.9992 - val_loss: 0.3437 - val_acc: 0.9364
Epoch 870/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0914 - acc: 0.9992 - val_loss: 0.3488 - val_acc: 0.9350
Epoch 871/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0913 - acc: 0.9991 - val_loss: 0.3473 - val_acc: 0.9367
Epoch 872/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0911 - acc: 0.9992 - val_loss: 0.3448 - val_acc: 0.9380
Epoch 873/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0907 - acc: 0.9993 - val_loss: 0.3439 - val_acc: 0.9373
Epoch 874/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0911 - acc: 0.9988 - val_loss: 0.3421 - val_acc: 0.9384
Epoch 875/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0904 - acc: 0.9992 - val_loss: 0.3430 - val_acc: 0.9365
Epoch 876/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0908 - acc: 0.9990 - val_loss: 0.3471 - val_acc: 0.9355
Epoch 877/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.0905 - acc: 0.9991 - val_loss: 0.3452 - val_acc: 0.9359
Epoch 878/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0905 - acc: 0.9990 - val_loss: 0.3379 - val_acc: 0.9372
Epoch 879/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0906 - acc: 0.9989 - val_loss: 0.3442 - val_acc: 0.9369
Epoch 880/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0903 - acc: 0.9990 - val_loss: 0.3413 - val_acc: 0.9363
Epoch 881/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0898 - acc: 0.9992 - val_loss: 0.3437 - val_acc: 0.9354
Epoch 882/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.0898 - acc: 0.9992 - val_loss: 0.3421 - val_acc: 0.9371
Epoch 883/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0897 - acc: 0.9991 - val_loss: 0.3442 - val_acc: 0.9363
Epoch 884/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0900 - acc: 0.9990 - val_loss: 0.3471 - val_acc: 0.9366
Epoch 885/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0897 - acc: 0.9991 - val_loss: 0.3443 - val_acc: 0.9361
Epoch 886/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0892 - acc: 0.9990 - val_loss: 0.3434 - val_acc: 0.9355
Epoch 887/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0890 - acc: 0.9991 - val_loss: 0.3411 - val_acc: 0.9367
Epoch 888/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0889 - acc: 0.9992 - val_loss: 0.3478 - val_acc: 0.9338
Epoch 889/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0889 - acc: 0.9991 - val_loss: 0.3404 - val_acc: 0.9366
Epoch 890/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0889 - acc: 0.9991 - val_loss: 0.3356 - val_acc: 0.9373
Epoch 891/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0886 - acc: 0.9992 - val_loss: 0.3358 - val_acc: 0.9362
Epoch 892/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0883 - acc: 0.9992 - val_loss: 0.3380 - val_acc: 0.9368
Epoch 893/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.0886 - acc: 0.9991 - val_loss: 0.3369 - val_acc: 0.9374
Epoch 894/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0881 - acc: 0.9993 - val_loss: 0.3397 - val_acc: 0.9386
Epoch 895/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0885 - acc: 0.9991 - val_loss: 0.3400 - val_acc: 0.9365
Epoch 896/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0883 - acc: 0.9989 - val_loss: 0.3367 - val_acc: 0.9355
Epoch 897/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0886 - acc: 0.9986 - val_loss: 0.3375 - val_acc: 0.9361
Epoch 898/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0878 - acc: 0.9989 - val_loss: 0.3444 - val_acc: 0.9354
Epoch 899/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0875 - acc: 0.9992 - val_loss: 0.3444 - val_acc: 0.9367
Epoch 900/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0877 - acc: 0.9990 - val_loss: 0.3457 - val_acc: 0.9353
Epoch 901/1000
lr changed to 9.999999310821295e-05
500/500 [==============================] - 66s 132ms/step - loss: 0.0873 - acc: 0.9992 - val_loss: 0.3442 - val_acc: 0.9350
Epoch 902/1000
500/500 [==============================] - 66s 133ms/step - loss: 0.0867 - acc: 0.9994 - val_loss: 0.3425 - val_acc: 0.9361
Epoch 903/1000
500/500 [==============================] - 66s 132ms/step - loss: 0.0874 - acc: 0.9991 - val_loss: 0.3432 - val_acc: 0.9358
Epoch 904/1000
500/500 [==============================] - 66s 131ms/step - loss: 0.0872 - acc: 0.9992 - val_loss: 0.3431 - val_acc: 0.9360
Epoch 905/1000
500/500 [==============================] - 66s 131ms/step - loss: 0.0871 - acc: 0.9991 - val_loss: 0.3426 - val_acc: 0.9371
Epoch 906/1000
500/500 [==============================] - 66s 132ms/step - loss: 0.0868 - acc: 0.9991 - val_loss: 0.3422 - val_acc: 0.9371
Epoch 907/1000
500/500 [==============================] - 66s 132ms/step - loss: 0.0869 - acc: 0.9993 - val_loss: 0.3418 - val_acc: 0.9368
Epoch 908/1000
500/500 [==============================] - 66s 132ms/step - loss: 0.0870 - acc: 0.9991 - val_loss: 0.3415 - val_acc: 0.9366
Epoch 909/1000
500/500 [==============================] - 66s 131ms/step - loss: 0.0863 - acc: 0.9995 - val_loss: 0.3410 - val_acc: 0.9371
Epoch 910/1000
500/500 [==============================] - 66s 131ms/step - loss: 0.0870 - acc: 0.9991 - val_loss: 0.3405 - val_acc: 0.9363
Epoch 911/1000
500/500 [==============================] - 66s 132ms/step - loss: 0.0864 - acc: 0.9995 - val_loss: 0.3412 - val_acc: 0.9367
Epoch 912/1000
500/500 [==============================] - 66s 132ms/step - loss: 0.0862 - acc: 0.9995 - val_loss: 0.3403 - val_acc: 0.9370
Epoch 913/1000
500/500 [==============================] - 78s 155ms/step - loss: 0.0862 - acc: 0.9995 - val_loss: 0.3399 - val_acc: 0.9368
Epoch 914/1000
500/500 [==============================] - 84s 168ms/step - loss: 0.0860 - acc: 0.9997 - val_loss: 0.3402 - val_acc: 0.9373
Epoch 915/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0865 - acc: 0.9994 - val_loss: 0.3403 - val_acc: 0.9371
Epoch 916/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0866 - acc: 0.9993 - val_loss: 0.3399 - val_acc: 0.9369
Epoch 917/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0868 - acc: 0.9992 - val_loss: 0.3385 - val_acc: 0.9378
Epoch 918/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.0865 - acc: 0.9993 - val_loss: 0.3374 - val_acc: 0.9376
Epoch 919/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0861 - acc: 0.9995 - val_loss: 0.3378 - val_acc: 0.9373
Epoch 920/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0864 - acc: 0.9993 - val_loss: 0.3373 - val_acc: 0.9380
Epoch 921/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0863 - acc: 0.9995 - val_loss: 0.3374 - val_acc: 0.9375
Epoch 922/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0863 - acc: 0.9993 - val_loss: 0.3371 - val_acc: 0.9376
Epoch 923/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.0861 - acc: 0.9995 - val_loss: 0.3372 - val_acc: 0.9370
Epoch 924/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0860 - acc: 0.9994 - val_loss: 0.3374 - val_acc: 0.9369
Epoch 925/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0860 - acc: 0.9996 - val_loss: 0.3375 - val_acc: 0.9368
Epoch 926/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0862 - acc: 0.9994 - val_loss: 0.3378 - val_acc: 0.9373
Epoch 927/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0864 - acc: 0.9992 - val_loss: 0.3384 - val_acc: 0.9371
Epoch 928/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0863 - acc: 0.9993 - val_loss: 0.3386 - val_acc: 0.9367
Epoch 929/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0861 - acc: 0.9995 - val_loss: 0.3392 - val_acc: 0.9365
Epoch 930/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0861 - acc: 0.9995 - val_loss: 0.3386 - val_acc: 0.9368
Epoch 931/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0861 - acc: 0.9995 - val_loss: 0.3384 - val_acc: 0.9375
Epoch 932/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0856 - acc: 0.9996 - val_loss: 0.3388 - val_acc: 0.9376
Epoch 933/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0859 - acc: 0.9995 - val_loss: 0.3390 - val_acc: 0.9376
Epoch 934/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0861 - acc: 0.9995 - val_loss: 0.3389 - val_acc: 0.9375
Epoch 935/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0859 - acc: 0.9994 - val_loss: 0.3390 - val_acc: 0.9376
Epoch 936/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0859 - acc: 0.9994 - val_loss: 0.3397 - val_acc: 0.9373
Epoch 937/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0858 - acc: 0.9995 - val_loss: 0.3396 - val_acc: 0.9371
Epoch 938/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0860 - acc: 0.9994 - val_loss: 0.3390 - val_acc: 0.9379
Epoch 939/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0859 - acc: 0.9993 - val_loss: 0.3393 - val_acc: 0.9382
Epoch 940/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0858 - acc: 0.9994 - val_loss: 0.3391 - val_acc: 0.9379
Epoch 941/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0858 - acc: 0.9995 - val_loss: 0.3392 - val_acc: 0.9378
Epoch 942/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0857 - acc: 0.9995 - val_loss: 0.3396 - val_acc: 0.9382
Epoch 943/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0858 - acc: 0.9995 - val_loss: 0.3403 - val_acc: 0.9376
Epoch 944/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0859 - acc: 0.9993 - val_loss: 0.3405 - val_acc: 0.9374
Epoch 945/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0854 - acc: 0.9996 - val_loss: 0.3402 - val_acc: 0.9371
Epoch 946/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0859 - acc: 0.9994 - val_loss: 0.3398 - val_acc: 0.9376
Epoch 947/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0857 - acc: 0.9994 - val_loss: 0.3397 - val_acc: 0.9371
Epoch 948/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.0855 - acc: 0.9996 - val_loss: 0.3396 - val_acc: 0.9375
Epoch 949/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0853 - acc: 0.9996 - val_loss: 0.3398 - val_acc: 0.9376
Epoch 950/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0856 - acc: 0.9996 - val_loss: 0.3397 - val_acc: 0.9378
Epoch 951/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0856 - acc: 0.9994 - val_loss: 0.3393 - val_acc: 0.9375
Epoch 952/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0857 - acc: 0.9996 - val_loss: 0.3397 - val_acc: 0.9374
Epoch 953/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0854 - acc: 0.9995 - val_loss: 0.3400 - val_acc: 0.9378
Epoch 954/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0856 - acc: 0.9995 - val_loss: 0.3401 - val_acc: 0.9368
Epoch 955/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.0855 - acc: 0.9994 - val_loss: 0.3403 - val_acc: 0.9370
Epoch 956/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0856 - acc: 0.9994 - val_loss: 0.3405 - val_acc: 0.9371
Epoch 957/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0857 - acc: 0.9994 - val_loss: 0.3408 - val_acc: 0.9375
Epoch 958/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0854 - acc: 0.9994 - val_loss: 0.3405 - val_acc: 0.9374
Epoch 959/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0856 - acc: 0.9993 - val_loss: 0.3408 - val_acc: 0.9375
Epoch 960/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0856 - acc: 0.9995 - val_loss: 0.3407 - val_acc: 0.9369
Epoch 961/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0853 - acc: 0.9996 - val_loss: 0.3402 - val_acc: 0.9371
Epoch 962/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.0855 - acc: 0.9994 - val_loss: 0.3399 - val_acc: 0.9371
Epoch 963/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.0852 - acc: 0.9996 - val_loss: 0.3400 - val_acc: 0.9378
Epoch 964/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.0856 - acc: 0.9994 - val_loss: 0.3399 - val_acc: 0.9375
Epoch 965/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0854 - acc: 0.9994 - val_loss: 0.3396 - val_acc: 0.9375
Epoch 966/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0852 - acc: 0.9995 - val_loss: 0.3391 - val_acc: 0.9368
Epoch 967/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0854 - acc: 0.9994 - val_loss: 0.3383 - val_acc: 0.9374
Epoch 968/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0854 - acc: 0.9994 - val_loss: 0.3384 - val_acc: 0.9375
Epoch 969/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0851 - acc: 0.9997 - val_loss: 0.3383 - val_acc: 0.9375
Epoch 970/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.0853 - acc: 0.9995 - val_loss: 0.3388 - val_acc: 0.9365
Epoch 971/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0851 - acc: 0.9996 - val_loss: 0.3381 - val_acc: 0.9356
Epoch 972/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0855 - acc: 0.9994 - val_loss: 0.3387 - val_acc: 0.9362
Epoch 973/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0857 - acc: 0.9994 - val_loss: 0.3385 - val_acc: 0.9372
Epoch 974/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0854 - acc: 0.9995 - val_loss: 0.3385 - val_acc: 0.9373
Epoch 975/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0851 - acc: 0.9996 - val_loss: 0.3380 - val_acc: 0.9375
Epoch 976/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0853 - acc: 0.9994 - val_loss: 0.3380 - val_acc: 0.9379
Epoch 977/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0854 - acc: 0.9994 - val_loss: 0.3374 - val_acc: 0.9376
Epoch 978/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0851 - acc: 0.9996 - val_loss: 0.3376 - val_acc: 0.9379
Epoch 979/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0853 - acc: 0.9995 - val_loss: 0.3380 - val_acc: 0.9378
Epoch 980/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0852 - acc: 0.9995 - val_loss: 0.3376 - val_acc: 0.9381
Epoch 981/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0854 - acc: 0.9995 - val_loss: 0.3377 - val_acc: 0.9381
Epoch 982/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.0849 - acc: 0.9996 - val_loss: 0.3373 - val_acc: 0.9384
Epoch 983/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0852 - acc: 0.9993 - val_loss: 0.3372 - val_acc: 0.9379
Epoch 984/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0848 - acc: 0.9997 - val_loss: 0.3368 - val_acc: 0.9381
Epoch 985/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0852 - acc: 0.9994 - val_loss: 0.3373 - val_acc: 0.9382
Epoch 986/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0847 - acc: 0.9997 - val_loss: 0.3372 - val_acc: 0.9380
Epoch 987/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0848 - acc: 0.9996 - val_loss: 0.3371 - val_acc: 0.9387
Epoch 988/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0853 - acc: 0.9993 - val_loss: 0.3377 - val_acc: 0.9380
Epoch 989/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0851 - acc: 0.9995 - val_loss: 0.3371 - val_acc: 0.9385
Epoch 990/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0848 - acc: 0.9996 - val_loss: 0.3379 - val_acc: 0.9384
Epoch 991/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0849 - acc: 0.9994 - val_loss: 0.3377 - val_acc: 0.9380
Epoch 992/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0852 - acc: 0.9993 - val_loss: 0.3370 - val_acc: 0.9381
Epoch 993/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0851 - acc: 0.9994 - val_loss: 0.3371 - val_acc: 0.9380
Epoch 994/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0848 - acc: 0.9996 - val_loss: 0.3371 - val_acc: 0.9381
Epoch 995/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0849 - acc: 0.9994 - val_loss: 0.3381 - val_acc: 0.9381
Epoch 996/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0849 - acc: 0.9996 - val_loss: 0.3379 - val_acc: 0.9379
Epoch 997/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0853 - acc: 0.9993 - val_loss: 0.3384 - val_acc: 0.9377
Epoch 998/1000
500/500 [==============================] - 65s 129ms/step - loss: 0.0849 - acc: 0.9995 - val_loss: 0.3393 - val_acc: 0.9369
Epoch 999/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0849 - acc: 0.9994 - val_loss: 0.3395 - val_acc: 0.9369
Epoch 1000/1000
500/500 [==============================] - 65s 130ms/step - loss: 0.0847 - acc: 0.9996 - val_loss: 0.3389 - val_acc: 0.9371
Train loss: 0.08910960255563259
Train accuracy: 0.9977200021743774
Test loss: 0.3388938118517399
Test accuracy: 0.9371000009775162测试准确率到了93.71%，比之前的都高一点。
Minghang Zhao, Shisheng Zhong, Xuyun Fu, Baoping Tang, Shaojiang Dong, Michael Pecht, Deep Residual Networks with Adaptively Parametric Rectifier Linear Units for Fault Diagnosis, IEEE Transactions on Industrial Electronics, 2020, DOI: 10.1109/TIE.2020.2972458
https://ieeexplore.ieee.org/document/8998530
————————————————
版权声明：本文为CSDN博主「dangqing1988」的原创文章，遵循 CC 4.0 BY-SA 版权协议，转载请附上原文出处链接及本声明。
原文链接：https://blog.csdn.net/dangqing1988/article/details/105688144
