如何查看哪些测试数据分类错误

Question

输出的准确率不是100%，因此存在网络错误分类的文本。我怎么看这些文字后，网络？你知道吗

import numpy as np
import keras
from keras.datasets import imdb
from keras.models import Sequential
from keras.layers import Dense, Activation,Dropout
from keras.preprocessing.text import Tokenizer
import matplotlib.pyplot as plt
from keras import optimizers
from keras.layers import Conv1D, GlobalMaxPooling1D

np.random.seed(42)

max_features = 10000
maxlen = 400
batch_size = 64
embedding_dims = 200
filters = 150
kernel_size = 5
hidden_dims = 50
epochs =5

(x_train, y_train), (x_test, y_test) = imdb.load_data(num_words=1000)

print(x_train.shape)
print(x_test.shape)
print(x_train[0])
print(y_train[0])
tokenizer = Tokenizer(num_words=1000)
x_train = tokenizer.sequences_to_matrix(x_train, mode='binary')
x_test = tokenizer.sequences_to_matrix(x_test, mode='binary')
print(x_train[0])

num_classes = 2
y_train = keras.utils.to_categorical(y_train, num_classes)
y_test = keras.utils.to_categorical(y_test, num_classes)
print(y_train.shape)
print(y_test.shape)

model = Sequential()

model.add(Dense(512,input_dim = 1000,activation = 'relu'))
model.add(Dropout(0.2))
model.add(Dense(128,activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(64,activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(32, activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(num_classes,activation='sigmoid'))
model.summary()



opt = optimizers.Adam(lr=0.1, beta_1=0.9, beta_2=0.999, epsilon=1e-08, decay=0.0)
model.compile(loss="binary_crossentropy", optimizer=opt, metrics=['accuracy'])

clf = model.fit(x_train, y_train, batch_size=128, epochs=5, validation_data=(x_test, y_test))

score = model.evaluate(x_test, y_test, verbose=1)
print("Accuracy: ", score[1])

我试过这个代码，但是出错了

y_pred = model.predict(x_test)

# bolean mask
mask = y_pred != y_test

# print rows that was classified incorrectly    
print(x_test[mask])

print(x_test[mask]) IndexError: boolean index did not match indexed array along dimension 1; dimension is 1000 but corresponding boolean dimension is 2

Answer 1

我修改了您的完整代码，使它只在一个类中运行（因为我们正在研究一个二进制问题），您可以研究错误分类的示例。结果证明，您使用的模型完全不适合您的任务。你知道吗

import numpy as np
import keras
from keras.datasets import imdb
from keras.models import Sequential
from keras.layers import Dense, Activation,Dropout
from keras.preprocessing.text import Tokenizer
import matplotlib.pyplot as plt
from keras import optimizers
from keras.layers import Conv1D, GlobalMaxPooling1D
import pandas as pd

np.random.seed(42)

max_features = 10000
maxlen = 400
batch_size = 64
embedding_dims = 200
filters = 150
kernel_size = 5
hidden_dims = 50
epochs =5

(x_train, y_train), (x_test, y_test) = imdb.load_data(num_words=1000)

print(x_train.shape)
print(x_test.shape)
print(x_train[0])
print(y_train[0])
tokenizer = Tokenizer(num_words=1000)
x_train = tokenizer.sequences_to_matrix(x_train, mode='binary')
x_test = tokenizer.sequences_to_matrix(x_test, mode='binary')
print(x_train[0])

num_classes = 1
#y_train = keras.utils.to_categorical(y_train, num_classes)
#y_test = keras.utils.to_categorical(y_test, num_classes)
print(y_train.shape)
print(y_test.shape)

model = Sequential()

model.add(Dense(512,input_dim = 1000,activation = 'relu'))
model.add(Dropout(0.2))
model.add(Dense(128,activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(64,activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(32, activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(num_classes,activation='sigmoid'))
model.summary()



opt = optimizers.Adam()
model.compile(loss="binary_crossentropy", optimizer=opt, metrics=['accuracy'])

clf = model.fit(x_train, y_train, batch_size=128, epochs=5, validation_data=(x_test, y_test))

score = model.evaluate(x_test, y_test, verbose=1)
print("Accuracy: ", score[1])

y_pred = model.predict(x_test)
df_test_pred = pd.concat([pd.DataFrame(x_test), pd.DataFrame(y_test, columns= ['test']), pd.DataFrame(y_pred, columns= ['pred'])], axis=1)

df_wrong= df_test_pred[df_test_pred['test'] != df_test_pred['pred']]