Keras模型的GridSearchCV：“功能”对象没有属性“ predict_classes”

问题描述

我建立了用于多标签分类的神经网络，效果很好。

1. 我的训练集特征是基因表达水平。他们是floats

2. 目标是与基因表达相关的分子途径。它们是二进制0/1。

3. 神经网络的预测是给定基因表达后分子途径被激活的可能性。

我的问题是，为了进行超参数调整，我正在使用sklearn.model_selection.gridsearchcv，但始终遇到上述错误。

以下是可复制的代码：

from sklearn.model_selection import gridsearchcv
from tensorflow.keras.wrappers.scikit_learn import KerasClassifier
import numpy as np
import pandas as pd
import tensorflow as tf

#some datas
train = np.random.random((10,20))
target = np.random.binomial(1,0.1,(10,5))

# Build the model
def create_model(): 
    inputs = tf.keras.Input(shape=(20,))
    x = tf.keras.layers.Dense(400,activation=tf.nn.relu)(inputs)
    outputs = tf.keras.layers.Dense(5,activation=tf.nn.sigmoid)(x)
    model = tf.keras.Model(inputs=inputs,outputs=outputs)
    model.compile(loss='binary_crossentropy',optimizer= 'Adam')
    return model

model = KerasClassifier(build_fn=create_model,verbose=1)

param_grid = {'epochs':[10,20],'batch_size':[200],}

gs = gridsearchcv(
    estimator=model,param_grid=param_grid,cv=3,n_jobs=-1,scoring= 'accuracy',verbose=2,)

fitted = gs.fit(train,target)

以下是由行fitted = gs.fit(train,target)

引起的错误

AttributeError: 'Functional' object has no attribute 'predict_classes'

有人可以给我一个线索吗？

解决方法

'Functional' object has no attribute 'predict_classes'的确如此。 'predict_classes'仅适用于Sequential模型。为了使代码正常工作，您需要将其调整为适用于多类proba预测，例如：

from sklearn.model_selection import GridSearchCV
from tensorflow.keras.wrappers.scikit_learn import KerasClassifier
import numpy as np
import pandas as pd
import tensorflow as tf

#some datas
train = np.random.random((10,20))
target = np.random.binomial(1,0.1,(10,5))

# Build the model
def create_model(): 
    inputs = tf.keras.Input(shape=(20,))
    x = tf.keras.layers.Dense(400,activation=tf.nn.relu)(inputs)
    outputs = tf.keras.layers.Dense(5,activation=tf.nn.softmax)(x)
    model = tf.keras.Model(inputs=inputs,outputs=outputs)
    model.compile(loss='categorical_crossentropy',optimizer= 'Adam')
    return model

model = KerasClassifier(build_fn=create_model,verbose=1)

param_grid = {'epochs':[10,20],'batch_size':[200],}

gs = GridSearchCV(
    estimator=model,param_grid=param_grid,cv=3,n_jobs=-1,verbose=2,)

fitted = gs.fit(train,target)

那你就可以了。

输出：

---

fitting 3 folds for each of 2 candidates,totalling 6 fits
[Parallel(n_jobs=-1)]: Using backend LokyBackend with 12 concurrent workers.
[Parallel(n_jobs=-1)]: Done   3 out of   6 | elapsed:    2.3s remaining:    2.3s
[Parallel(n_jobs=-1)]: Done   6 out of   6 | elapsed:    2.3s finished
Epoch 1/10
1/1 [==============================] - 0s 191ms/step - loss: 1.5599
Epoch 2/10
1/1 [==============================] - 0s 2ms/step - loss: 1.5250
Epoch 3/10
1/1 [==============================] - 0s 2ms/step - loss: 1.4932
Epoch 4/10
1/1 [==============================] - 0s 2ms/step - loss: 1.4649
Epoch 5/10
1/1 [==============================] - 0s 2ms/step - loss: 1.4396
Epoch 6/10
1/1 [==============================] - 0s 2ms/step - loss: 1.4165
Epoch 7/10
1/1 [==============================] - 0s 2ms/step - loss: 1.3950
Epoch 8/10
1/1 [==============================] - 0s 2ms/step - loss: 1.3746
Epoch 9/10
1/1 [==============================] - 0s 2ms/step - loss: 1.3553
Epoch 10/10
1/1 [==============================] - 0s 2ms/step - loss: 1.3370