keras用plot_model函数画模型的结构图，保存为图片

anaconda中运行时出现

C:\Program Files (x86)\Microsoft Visual Studio\Shared\Anaconda3_64\lib\site-packages\sklearn\metrics\classification.py:1135: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 due to no predicted samples.
  'precision', 'predicted', average, warn_for)

ps：可能第二次运行的时候，警告就消失了
参考链接：https://www.itranslater.com/qa/details/2582485825728545792

参考链接：https://blog.csdn.net/So_that/article/details/89787366

CNN——Flatten layer
参考链接：https://www.jianshu.com/p/32756e5dd969

keras用plot_model函数画模型的结构图，保存为图片
首先是model.summary()
lstm_test_model.summary(line_length=200,positions=[0.30,0.60,0.7,1.0])
plot_model()保存模型结构图

#导入下面的库
from tensorflow.keras.utils import plot_model  
import pydotplus  
#参数 ：模型名称，结构图保存位置，是否展示shape
plot_model(lstm_test_model,to_file='lstm_test_model.png',show_shapes=True)

含有参数的模型函数的结果保存

代码：

# w2v_model=Word2Vec.load('sentiment_analysis/w2v_model.pkl')
# # 预训练的词向量中没有出现的词用0向量表示
# embedding_matrix = np.zeros((len(vocab) + 1, 300))
# for word, i in vocab.items():
#     try:
#         embedding_vector = w2v_model[str(word)]
#         embedding_matrix[i] = embedding_vector
#     except KeyError:
#         continue
# 构建TextCNN模型
def TextCNN_model_2(x_train_padded_seqs, y_train, x_test_padded_seqs, y_test, embedding_matrix):
    # 模型结构：词嵌入-卷积池化*3-拼接-全连接-dropout-全连接
    main_input = Input(shape=(3397,), dtype='float64')
    # 词嵌入（使用预训练的词向量）
    # embedder = Embedding(len(word2idx) + 1, 300, input_length=19422, weights=[embedding_matrix], trainable=False)
    # 之前训练word2vec时，停用词已经加过1了，因此这里不再加了，将上面的一句改为下面的一句
    embedder = Embedding(len(word2idx), 300, input_length=3397, weights=[embedding_matrix], trainable=False)
    # embedder = Embedding(len(vocab) + 1, 300, input_length=50, trainable=False)
    #     print(embedding_matrix[1])            #与前面训练好的embedding_matrix[1]结果相同
    embed = embedder(main_input)
    # 词窗大小分别为3,4,5
    cnn1 = Conv1D(256, 3, padding='same', strides=1, activation='relu')(embed)
    cnn1 = MaxPooling1D(pool_size=38)(cnn1)
    cnn2 = Conv1D(256, 4, padding='same', strides=1, activation='relu')(embed)
    cnn2 = MaxPooling1D(pool_size=37)(cnn2)
    cnn3 = Conv1D(256, 5, padding='same', strides=1, activation='relu')(embed)
    cnn3 = MaxPooling1D(pool_size=36)(cnn3)
    # 合并三个模型的输出向量
    cnn = concatenate([cnn1, cnn2, cnn3], axis=1)
    flat = Flatten()(cnn)
    drop = Dropout(0.2)(flat)
    main_output = Dense(9, activation='softmax')(drop)
    model = Model(inputs=main_input, outputs=main_output)
    model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])
    model.fit(x_train_padded_seqs, y_train, batch_size=2, epochs=2)
    result = model.predict(x_test_padded_seqs)  # 预测样本属于每个类别的概率
    result_labels = np.argmax(result, axis=1)  # 获得最大概率对应的标签
    y_predict = result_labels
    y_tests = np.argmax(y_test, axis=1)
    print('准确率', metrics.accuracy_score(y_tests, y_predict))
    print('平均f1-score:', metrics.f1_score(y_tests, y_predict, average='weighted'))

修改之后的代码：

from keras.layers import Conv1D, MaxPool1D, Dense, Flatten, concatenate, Embedding
from keras.layers import MaxPooling1D,normalization
from sklearn import metrics
from keras.models import Model
# 构建TextCNN模型
def TextCNN_model_2(embedding_matrix):
    # 模型结构：词嵌入-卷积池化*3-拼接-全连接-dropout-全连接
    main_input = Input(shape=(19422,), dtype='float64')
    # 词嵌入（使用预训练的词向量）
    embedder = Embedding(len(word2idx), 300, input_length=19422, weights=[embedding_matrix], trainable=False)
    embed = embedder(main_input)
    # 词窗大小分别为3,4,5
    cnn1 = Conv1D(256, 3, padding='same', strides=1, activation='relu')(embed)
    cnn1 = MaxPooling1D(pool_size=38)(cnn1)
    cnn2 = Conv1D(256, 4, padding='same', strides=1, activation='relu')(embed)
    cnn2 = MaxPooling1D(pool_size=37)(cnn2)
    cnn3 = Conv1D(256, 5, padding='same', strides=1, activation='relu')(embed)
    cnn3 = MaxPooling1D(pool_size=36)(cnn3)
    # 合并三个模型的输出向量
    cnn = concatenate([cnn1, cnn2, cnn3], axis=1)
    flat = Flatten()(cnn)
    drop = Dropout(0.2)(flat)
    main_output = Dense(3, activation='softmax')(drop)
    model = Model(inputs=main_input, outputs=main_output)
    model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])
    print(model.summary())
    return model

然后再是：

model = TextCNN_model_2(embedMatrix)
model.fit(X_tra_idx, y_tra_idx, validation_data=(X_val_idx, y_val_idx),
          epochs=3, batch_size=10, verbose=1)

#导入下面的库
from tensorflow.keras.utils import plot_model  
import pydotplus  
#参数 ：模型名称，结构图保存位置，是否展示shape
plot_model(model,to_file='textCNN_model.png',show_shapes=True)

即可得到保存的textCNN模型结构图