我是tf和keras的新手，我正在使用带有2.2.4-tf tensorflow.keras的colab笔记本python 3。
呼叫pip list显示
imbalanced-learn 0.4.3
imblearn 0.0
我正在尝试使用imblearn.keras BalancedBatchGenerator，然后使用model.fit_generator并出现属性错误
AttributeError: 'BalancedBatchGenerator' object has no attribute 'shape'
输入的训练数据和标签是具有shape属性的numpy数组，因此不确定出了什么问题
nest.flatten(Xfeatures_train)[0]
>>array([[9.97292995e-03, 4.43795400e-09, 1.99458601e-08, ...,
1.99458599e-02, 5.98375825e-03, 9.97293055e-01],
[9.97279119e-03, 2.79238166e-09, 5.98367507e-08, ...,
[5.35506010e-02, 2.52044821e-08, 7.14007982e-08, ...,
7.14007989e-02, 2.14202404e-02, 9.63910818e-01]],
dtype=float32)
>> (1106, 7)
nest.flatten(ylabels_train)[0]
>>array([[0., 0., 1.],
[0., 1., 0.],
[0., 1., 0.],
...,
[0., 0., 1.],
[0., 1., 0.],
[0., 0., 1.]], dtype=float32)
>> (1106, 3)
feature_dimensions=Xfeatures_train.shape[1]
tensor_dimensions=7
n_hidden_1 = 256 # 1st hidden layer
n_hidden_2 = 128
n_classes = 3 # classes for prediction
model = tf.keras.Sequential([
tf.keras.layers.Dense(tensor_dimensions, input_dim=feature_dimensions, activation='relu'),
tf.keras.layers.Dense(n_hidden_1, activation='relu'),
tf.keras.layers.Dropout(0.1),
tf.keras.layers.Dense(n_hidden_2, activation='relu'),
tf.keras.layers.Dropout(0.2),
tf.keras.layers.Dense(n_classes, activation='softmax')
])
model.compile(optimizer='sgd', loss='categorical_crossentropy',
metrics=['accuracy'])
from imblearn.keras import BalancedBatchGenerator
from imblearn.under_sampling import RandomUnderSampler
training_generator=BalancedBatchGenerator(Xfeatures_train, ylabels_train, sampler=RandomUnderSampler(), batch_size=10)
training_epochs = 2
callback_history = model.fit_generator(training_generator, epochs=training_epochs, steps_per_epoch=10, verbose=2)
ERROR ----
----> 3 callback_history = model.fit_generator(training_generator, epochs=training_epochs, steps_per_epoch=10, verbose=2)
4 #history = model.fit_generator(BalancedBatchGenerator(Xfeatures_train, ylabels_train, sampler=RandomUnderSampler(), batch_size=batch_size, random_state=42),
5 # steps_per_epoch=len(Xfeatures_train)//batch_size,
2 frames
/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/training.py in fit_generator(self, generator, steps_per_epoch, epochs, verbose, callbacks, validation_data, validation_steps, class_weight, max_queue_size, workers, use_multiprocessing, shuffle, initial_epoch)
1424 use_multiprocessing=use_multiprocessing,
1425 shuffle=shuffle,
-> 1426 initial_epoch=initial_epoch)
1427
1428 def evaluate_generator(self,
/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/training_generator.py in model_iteration(model, data, steps_per_epoch, epochs, verbose, callbacks, validation_data, validation_steps, class_weight, max_queue_size, workers, use_multiprocessing, shuffle, initial_epoch, mode, batch_size, **kwargs)
113 batch_size=batch_size,
114 epochs=epochs - initial_epoch,
--> 115 shuffle=shuffle)
116
117 do_validation = validation_data is not None
/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/training_generator.py in convert_to_generator_like(data, batch_size, steps_per_epoch, epochs, shuffle)
375
376 # Create generator from NumPy or EagerTensor Input.
--> 377 num_samples = int(nest.flatten(data)[0].shape[0])
378 if batch_size is None:
379 raise ValueError('You must specify `batch_size`')
AttributeError: 'BalancedBatchGenerator' object has no attribute 'shape'
我可以使用以下代码重现此错误
from sklearn.datasets import load_iris
iris = load_iris()
from imblearn.datasets import make_imbalance
class_dict = dict()
class_dict[0] = 30; class_dict[1] = 50; class_dict[2] = 40
X, y = make_imbalance(iris.data, iris.target, class_dict)
X=X.astype('float32')
#import keras
y = tf.keras.utils.to_categorical(y, 3)
model1 = tf.keras.models.Sequential()
model1.add(tf.keras.layers.Dense(y.shape[1], input_dim=X.shape[1],
activation='softmax'))
model1.compile(optimizer='sgd', loss='categorical_crossentropy',
metrics=['accuracy'])
from imblearn.keras import BalancedBatchGenerator
from imblearn.under_sampling import NearMiss
from imblearn.under_sampling import RandomUnderSampler
training_generator1 = BalancedBatchGenerator(X, y, sampler=RandomUnderSampler(), batch_size=10, random_state=42)
#training_generator1
callback_history = model1.fit_generator(generator=training_generator1,
epochs=10, verbose=2)
在上面的代码中，如果我取消注释#import keras并从tf.keras更改为keras，则错误类似
看起来keras和tf.keras API有所不同
答案 0 :(得分：0)
此代码中的问题可能来自代码中的这一行。
export const processUserImage = async function(userUid: string, filename: string, imageUid: string, entityUid: string): Promise<any> {
if (!filename || !imageUid || !entityUid) {
return Promise.reject('Can\'t upload the image. Bad/missing datas');
}
const uploadedFilePath = `entities/uploads/${filename}`;
const image = bucket.file(uploadedFilePath);
const tmpLocalFilePath = path.join(os.tmpdir(), filename); // tmp/GcFuMyKvHzQKbKxLMvyO.jpg
await image.download({
destination: tmpLocalFilePath,
validation: false
});
const [ metadata, tmpLocalOptFilePath ] = await Promise.all([
sharp(tmpLocalFilePath).metadata(),
optimizeImageTo(tmpLocalFilePath, `entities/${entityUid}/${imageUid}.jpg`),
generateThumbnail(tmpLocalFilePath, `entities/${entityUid}/${imageUid}_${ImageFormat.getSlug(ImageFormat.THUMB)}.jpg`),
image.delete() // We can delete the uploaded tmp file, we're gonna work with local version now and reupload afterwards
]);
fs.unlinkSync(tmpLocalFilePath);
fs.unlinkSync(tmpLocalOptFilePath);
return admin.firestore().runTransaction(async(t) => {
const entity = await getEntity(entityUid, t);
if (entity.userUid !== userUid) return;
const entityRef = admin.firestore().doc(`entities/${entityUid}`);
t.set(entityRef, {
images: admin.firestore.FieldValue.arrayUnion({
uid: imageUid,
imageRatio: metadata.height * 100 / metadata.width,
color: { blue: 255, green: 255, red: 255 }
})
}, { merge: true });
});
}
如果您正在使用tensorflow，则将上面的行替换为
from imblearn.keras import BalancedBatchGenerator