def show_images(images, labels, preds):
plt.figure(figsize=(8, 4))
for i, image in enumerate(images):
plt.subplot(1, 6, i + 1, xticks=[], yticks=[])
image = image.numpy().transpose((1, 2, 0))
mean = np.array([0.485, 0.456, 0.406])
std = np.array([0.229, 0.224, 0.225])
image = image * std + mean
image = np.clip(image, 0., 1.)
plt.imshow(image)
col = 'green'
if preds[i] != labels[i]:
col = 'red'
plt.xlabel(f'{class_names[int(labels[i].numpy())]}')
plt.ylabel(f'{class_names[int(preds[i].numpy())]}', color=col)
plt.tight_layout()
plt.show()
if torch.cuda.is_available():
model.cuda()
for epoch in range(epoches): #循环epoch，设置学习率
if epoch in [epoches * 0.25, epoches * 0.5]:
for param_group in optimizier.param_groups:
param_group['lr'] *= 0.1
for img, _ in train_data:
img = img.view(img.size(0), -1)
img = Variable(img.cuda())
#forward
_, output = model(img) #把数据放入模型
loss = criterion(output, img) #损失
#backward
optimizier.zero_grad()
loss.backward()
optimizier.step()
print('epoch=', epoch, loss.data.float())
for param_group in optimizier.param_groups:
print(param_group['lr'])
if (epoch + 1) % 5 == 0:
pic = to_img(output.cpu().data)
if not os.path.exists('./simple_autoencoder'): #判断该路径是否存在
os.mkdir(pic, './simple_autoencoder/image_{}.png'.format(
epoch + 1)) #如果不存在则将处理后的数据放入该路径
code = Variable(torch.FloatTensor([1.19, -3.36, 2.06]).cuda())
decode = model.decoder(code) #解码
decode_img = to_img(decode).squeeze()
decode_img = decode_img.data.cpu().numpy() * 255
plt.imshow(decode_img.astype('uint8'), cmap='gray')
plt.show()
sys.path.append('.') # noqa: E402
from src.recurrent_attention_network_paper.StanfordDogLoader import StanfordDogLoader
if __name__ == '__main__':
# trainset=CUB200_loader('external/CUB_200_2011',split='train')
trainset = StanfordDogLoader('external/StanfordDog', split='train')
trainloader = torch.utils.data.DataLoader(trainset, batch_size=4, shuffle=True, collate_fn=trainset.CUB_collate,
num_workers=4)
first,labels= next(iter(trainloader))
print(first[0].mean())
print(first[0].std())
firstImage= trainset.tensor_to_img(first[0])
plt.imshow(firstImage)
plt.savefig('test.jpg')
# num = len(trainset)
#
# imageNum=10
#
# total_sum = 0
# dataIter= iter(trainloader)
# for i in range(0,imageNum):
# curr = next(dataIter)
# total_sum += curr[0].mean()
#
# mean = total_sum / imageNum
# print(mean)
epochs = 15
batch_size = 128
history = model.fit(x_train, x_train,
batch_size=batch_size,
epochs=epochs,
verbose=1,
validation_data=(x_test, x_test))
#-----------------------------------------------------查看解码效果--------------------------------------------
decoded_imgs = model.predict(x_test)
n = 10
plt.figure(figsize=(20, 6))
for i in range(n):
# 原图
ax = plt.subplot(3, n, i+1)
plt.imshow(x_test[i].reshape(28, 28))
plt.gray()
ax.get_xaxis().set_visible(False)
ax.get_yaxis().set_visible(False)
# 解码效果图
ax = plt.subplot(3, n, i+n+1)
plt.imshow(decoded_imgs[i].reshape(28, 28))
plt.gray()
ax.get_xaxis().set_visible(False)
ax.get_yaxis().set_visible(False)
plt.show()
#----------------------------------------------------训练过程可视化---------------------------------------------
print(history.history.keys())
model.add(Dropout(0.25))
model.add(Flatten())
model.add(Dense(1000, activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(num_classes, activation='softmax'))
model.summary()
model.compile(loss='categorical_crossentropy',
optimizer='adam',
metrics=['accuracy'])
hist = model.fit(x_train,
y_train,
batch_size=batch_size,
epochs=epochs,
verbose=1,
validation_data=(x_test, y_test))
score = model.evaluate(x_test, y_test, verbose=0)
print('Test loss:', score[0])
print('Test accuracy:', score[1])
print(
"====================================================================================================="
)
n = 0
plt.imshow(x_test[n].reshape(28, 28), cmap='Greys', interpolation='nearest')
plt.show()
print('The Answer is ', model.predict_classes(x_test[n].reshape(
(1, 28, 28, 1))))
epochs = 5
batch_size = 128
history = autoencoder.fit(x_train, x_train, batch_size=batch_size, epochs=epochs, verbose=1,validation_data=(x_test, x_test))
#----------------------------------------------------模型可视化---------------------------------
SVG(model_to_dot(autoencoder).create(prog='dot',format='svg'))
#----------------------------------------------------查看自编码器的压缩效果------------------------
#只取编码器做模型（取输入层x和隐藏层h，作为网络结构）
encoded_imgs = model.predict(x_test)
#打印10张测试集手写体的压缩效果
n = 10
plt.figure(figsize=(20, 8))
for i in range(n):
ax = plt.subplot(1, n, i+1)
plt.imshow(encoded_imgs[i].reshape(4, 16).T) #将8*8的特征转化为4*16的图像
plt.gray()
ax.get_xaxis().set_visible(False)
ax.get_yaxis().set_visible(False)
plt.show()
#----------------------------------------------------查看自编码器的解码效果-------------------------
decoded_imgs = autoencoder.predict(x_test) #decoded_imgs为输出层的效果
n = 10
plt.figure(figsize=(20, 6))
for i in range(n):
#打印原图
ax = plt.subplot(3, n, i+1)
plt.imshow(x_test[i].reshape(28, 28))
plt.gray()
ax.get_xaxis().set_visible(False)
rc('font', family='NanumBarunGothic')
wordcloud = WordCloud(
font_path='/content/CookieRun Regular.ttf', # 맥에선 한글폰트 설정 잘해야함.
background_color='white', # 배경 색깔 정하기
colormap='Accent_r', # 폰트 색깔 정하기
width=800,
height=800)
wordcloud_words = wordcloud.generate_from_frequencies(words)
array = wordcloud.to_array()
print(type(array)) # numpy.ndarray
print(array.shape) # (800, 800, 3)
fig = plt.figure(figsize=(10, 10))
plt.imshow(array, interpolation="bilinear")
plt.axis('off')
plt.show()
fig.savefig('business_anlytics_worldcloud.png')
for sentence in lists:
morphs.append(twitter.pos(sentence))
print(morphs)
noun_adj_adv_list = []
for sentence in morphs:
for word, tag in sentence:
if tag in ['Noun'] and ("것" not in word) and ("내" not in word) and (
"나" not in word) and ("수" not in word) and (
"게" not in word) and ("말" not in word):
noun_adj_adv_list.append(word)
history = autoencoder.fit(x_train, x_train,
batch_size=batch_size,
epochs=epochs,
verbose=1,
validation_data=(x_test, x_test))
#------------------------------------------查看自编码器的压缩效果-----------------------------
#为隐藏层的结果（encoder的最后一层），只取编码器做模型
conv_encoder = Model(x, h)
encoded_imgs = autoencoder.predict(x_test)
#打印10张测试集手写体的压缩效果
n = 10
plt.figure(figsize=(20, 8))
for i in range(n):
ax = plt.subplot(3, n, i+n+1)
plt.imshow(encoded_imgs[i].reshape(4, 16).T)
plt.gray()
ax.get_xaxis().set_visible(False)
plt.gray()
ax.get_xaxis().set_visible(False)
ax.get_yaxis().set_visible(False)
plt.show()
#------------------------------------------查看自编码器的解码效果------------------------------
decoded_imgs = autoencoder.predict(x_test)
n = 10
plt.figure(figsize=(20,6))
for i in range(n):
#原图
ax = plt.subplot(3, n, i+1)
plt.imshow(x_test[i].reshape(28, 28))
