我在keras中构建了一个多层Perceptron模型。但是我在训练过程中遇到了一个问题：准确度从第二个时期开始变得恒定并且等于1。我试图破坏激活功能和时代数量等但是徒劳无功。
这是脚本：
checkpointer4 = ModelCheckpoint(filepath="modelsup.h5",
verbose=0,
save_best_only=True,
save_weights_only=True)
model = Sequential()
model.add(Dense(64, input_dim=10, activation='relu'))
model.add(Dense(64, activation='relu'))
model.add(Dense(1, activation='sigmoid'))
model.compile(loss='binary_crossentropy',
optimizer='adam',
metrics=['accuracy'])
history = model.fit(x_train, y_train,
epochs=25,
batch_size=10,
validation_data=(x_test, y_test),
shuffle = True,
verbose=1,
callbacks=[checkpointer4]).history
model.save_weights("modelsup.h5")
model.load_weights("modelsup.h5")
score = model.evaluate(x_test, y_test, batch_size=25)
print('Test score:', score)
print(model.summary())
以下是结果的一部分：
Epoch 25/25
10/10000 [..............................] - ETA: 6s - loss: 1.1345e-07 - acc: 1.0000
120/10000 [..............................] - ETA: 5s - loss: 1.1425e-07 - acc: 1.0000
230/10000 [..............................] - ETA: 5s - loss: 1.1370e-07 - acc: 1.0000
320/10000 [..............................] - ETA: 5s - loss: 1.1339e-07 - acc: 1.0000
390/10000 [>.............................] - ETA: 5s - loss: 1.1325e-07 - acc: 1.0000
480/10000 [>.............................] - ETA: 5s - loss: 1.1353e-07 - acc: 1.0000
520/10000 [>.............................] - ETA: 6s - loss: 1.1367e-07 - acc: 1.0000
580/10000 [>.............................] - ETA: 6s - loss: 1.1358e-07 - acc: 1.0000
640/10000 [>.............................] - ETA: 6s - loss: 1.1357e-07 - acc: 1.0000
700/10000 [=>............................] - ETA: 7s - loss: 1.1369e-07 - acc: 1.0000
760/10000 [=>............................] - ETA: 7s - loss: 1.1380e-07 - acc: 1.0000
850/10000 [=>............................] - ETA: 6s - loss: 1.1388e-07 - acc: 1.0000
920/10000 [=>............................] - ETA: 6s - loss: 1.1389e-07 - acc: 1.0000
1000/10000 [==>...........................] - ETA: 6s - loss: 1.1391e-07 - acc: 1.0000
1060/10000 [==>...........................] - ETA: 6s - loss: 1.1383e-07 - acc: 1.0000
1140/10000 [==>...........................] - ETA: 6s - loss: 1.1373e-07 - acc: 1.0000
1230/10000 [==>...........................] - ETA: 6s - loss: 1.1374e-07 - acc: 1.0000
1310/10000 [==>...........................] - ETA: 6s - loss: 1.1378e-07 - acc: 1.0000
1380/10000 [===>..........................] - ETA: 6s - loss: 1.1382e-07 - acc: 1.0000
1480/10000 [===>..........................] - ETA: 6s - loss: 1.1375e-07 - acc: 1.0000
1540/10000 [===>..........................] - ETA: 6s - loss: 1.1371e-07 - acc: 1.0000
1620/10000 [===>..........................] - ETA: 6s - loss: 1.1379e-07 - acc: 1.0000
1720/10000 [====>.........................] - ETA: 6s - loss: 1.1383e-07 - acc: 1.0000
1800/10000 [====>.........................] - ETA: 6s - loss: 1.1383e-07 - acc: 1.0000
1890/10000 [====>.........................] - ETA: 6s - loss: 1.1389e-07 - acc: 1.0000
1940/10000 [====>.........................] - ETA: 6s - loss: 1.1391e-07 - acc: 1.0000
2020/10000 [=====>........................] - ETA: 6s - loss: 1.1389e-07 - acc: 1.0000
2100/10000 [=====>........................] - ETA: 6s - loss: 1.1389e-07 - acc: 1.0000
2160/10000 [=====>........................] - ETA: 6s - loss: 1.1392e-07 - acc: 1.0000
2240/10000 [=====>........................] - ETA: 5s - loss: 1.1388e-07 - acc: 1.0000
2320/10000 [=====>........................] - ETA: 5s - loss: 1.1385e-07 - acc: 1.0000
2390/10000 [======>.......................] - ETA: 5s - loss: 1.1386e-07 - acc: 1.0000
2460/10000 [======>.......................] - ETA: 5s - loss: 1.1391e-07 - acc: 1.0000
2580/10000 [======>.......................] - ETA: 5s - loss: 1.1393e-07 - acc: 1.0000
2660/10000 [======>.......................] - ETA: 5s - loss: 1.1390e-07 - acc: 1.0000
2770/10000 [=======>......................] - ETA: 5s - loss: 1.1394e-07 - acc: 1.0000
2870/10000 [=======>......................] - ETA: 5s - loss: 1.1396e-07 - acc: 1.0000
2970/10000 [=======>......................] - ETA: 5s - loss: 1.1397e-07 - acc: 1.0000
3030/10000 [========>.....................] - ETA: 5s - loss: 1.1396e-07 - acc: 1.0000
3140/10000 [========>.....................] - ETA: 4s - loss: 1.1402e-07 - acc: 1.0000
3210/10000 [========>.....................] - ETA: 4s - loss: 1.1403e-07 - acc: 1.0000
3320/10000 [========>.....................] - ETA: 4s - loss: 1.1403e-07 - acc: 1.0000
3410/10000 [=========>....................] - ETA: 4s - loss: 1.1403e-07 - acc: 1.0000
3520/10000 [=========>....................] - ETA: 4s - loss: 1.1405e-07 - acc: 1.0000
3610/10000 [=========>....................] - ETA: 4s - loss: 1.1402e-07 - acc: 1.0000
3700/10000 [==========>...................] - ETA: 4s - loss: 1.1402e-07 - acc: 1.0000
3840/10000 [==========>...................] - ETA: 4s - loss: 1.1401e-07 - acc: 1.0000
3940/10000 [==========>...................] - ETA: 4s - loss: 1.1401e-07 - acc: 1.0000
4040/10000 [===========>..................] - ETA: 4s - loss: 1.1401e-07 - acc: 1.0000
4160/10000 [===========>..................] - ETA: 3s - loss: 1.1407e-07 - acc: 1.0000
4270/10000 [===========>..................] - ETA: 3s - loss: 1.1410e-07 - acc: 1.0000
4390/10000 [============>.................] - ETA: 3s - loss: 1.1410e-07 - acc: 1.0000
4510/10000 [============>.................] - ETA: 3s - loss: 1.1411e-07 - acc: 1.0000
4650/10000 [============>.................] - ETA: 3s - loss: 1.1414e-07 - acc: 1.0000
4750/10000 [=============>................] - ETA: 3s - loss: 1.1414e-07 - acc: 1.0000
4850/10000 [=============>................] - ETA: 3s - loss: 1.1414e-07 - acc: 1.0000
4960/10000 [=============>................] - ETA: 3s - loss: 1.1417e-07 - acc: 1.0000
5110/10000 [==============>...............] - ETA: 3s - loss: 1.1415e-07 - acc: 1.0000
5190/10000 [==============>...............] - ETA: 3s - loss: 1.1414e-07 - acc: 1.0000
5340/10000 [===============>..............] - ETA: 2s - loss: 1.1417e-07 - acc: 1.0000
5440/10000 [===============>..............] - ETA: 2s - loss: 1.1417e-07 - acc: 1.0000
5550/10000 [===============>..............] - ETA: 2s - loss: 1.1418e-07 - acc: 1.0000
5690/10000 [================>.............] - ETA: 2s - loss: 1.1421e-07 - acc: 1.0000
5800/10000 [================>.............] - ETA: 2s - loss: 1.1421e-07 - acc: 1.0000
5920/10000 [================>.............] - ETA: 2s - loss: 1.1420e-07 - acc: 1.0000
6000/10000 [=================>............] - ETA: 2s - loss: 1.1420e-07 - acc: 1.0000
6100/10000 [=================>............] - ETA: 2s - loss: 1.1420e-07 - acc: 1.0000
6200/10000 [=================>............] - ETA: 2s - loss: 1.1421e-07 - acc: 1.0000
6300/10000 [=================>............] - ETA: 2s - loss: 1.1420e-07 - acc: 1.0000
6360/10000 [==================>...........] - ETA: 2s - loss: 1.1420e-07 - acc: 1.0000
6450/10000 [==================>...........] - ETA: 2s - loss: 1.1421e-07 - acc: 1.0000
6540/10000 [==================>...........] - ETA: 2s - loss: 1.1420e-07 - acc: 1.0000
6640/10000 [==================>...........] - ETA: 2s - loss: 1.1421e-07 - acc: 1.0000
6740/10000 [===================>..........] - ETA: 2s - loss: 1.1421e-07 - acc: 1.0000
6830/10000 [===================>..........] - ETA: 1s - loss: 1.1422e-07 - acc: 1.0000
6920/10000 [===================>..........] - ETA: 1s - loss: 1.1423e-07 - acc: 1.0000
7030/10000 [====================>.........] - ETA: 1s - loss: 1.1422e-07 - acc: 1.0000
7150/10000 [====================>.........] - ETA: 1s - loss: 1.1423e-07 - acc: 1.0000
7280/10000 [====================>.........] - ETA: 1s - loss: 1.1423e-07 - acc: 1.0000
7360/10000 [=====================>........] - ETA: 1s - loss: 1.1425e-07 - acc: 1.0000
7510/10000 [=====================>........] - ETA: 1s - loss: 1.1424e-07 - acc: 1.0000
7630/10000 [=====================>........] - ETA: 1s - loss: 1.1424e-07 - acc: 1.0000
7730/10000 [======================>.......] - ETA: 1s - loss: 1.1424e-07 - acc: 1.0000
7880/10000 [======================>.......] - ETA: 1s - loss: 1.1424e-07 - acc: 1.0000
8000/10000 [=======================>......] - ETA: 1s - loss: 1.1421e-07 - acc: 1.0000
8110/10000 [=======================>......] - ETA: 1s - loss: 1.1421e-07 - acc: 1.0000
8200/10000 [=======================>......] - ETA: 1s - loss: 1.1421e-07 - acc: 1.0000
8300/10000 [=======================>......] - ETA: 1s - loss: 1.1421e-07 - acc: 1.0000
8420/10000 [========================>.....] - ETA: 0s - loss: 1.1421e-07 - acc: 1.0000
8540/10000 [========================>.....] - ETA: 0s - loss: 1.1421e-07 - acc: 1.0000
8620/10000 [========================>.....] - ETA: 0s - loss: 1.1422e-07 - acc: 1.0000
8680/10000 [=========================>....] - ETA: 0s - loss: 1.1423e-07 - acc: 1.0000
8790/10000 [=========================>....] - ETA: 0s - loss: 1.1424e-07 - acc: 1.0000
8900/10000 [=========================>....] - ETA: 0s - loss: 1.1424e-07 - acc: 1.0000
9020/10000 [==========================>...] - ETA: 0s - loss: 1.1424e-07 - acc: 1.0000
9130/10000 [==========================>...] - ETA: 0s - loss: 1.1422e-07 - acc: 1.0000
9260/10000 [==========================>...] - ETA: 0s - loss: 1.1423e-07 - acc: 1.0000
9370/10000 [===========================>..] - ETA: 0s - loss: 1.1422e-07 - acc: 1.0000
9470/10000 [===========================>..] - ETA: 0s - loss: 1.1424e-07 - acc: 1.0000
9510/10000 [===========================>..] - ETA: 0s - loss: 1.1423e-07 - acc: 1.0000
9560/10000 [===========================>..] - ETA: 0s - loss: 1.1423e-07 - acc: 1.0000
9630/10000 [===========================>..] - ETA: 0s - loss: 1.1422e-07 - acc: 1.0000
9700/10000 [============================>.] - ETA: 0s - loss: 1.1423e-07 - acc: 1.0000
9830/10000 [============================>.] - ETA: 0s - loss: 1.1423e-07 - acc: 1.0000
9910/10000 [============================>.] - ETA: 0s - loss: 1.1424e-07 - acc: 1.0000
9990/10000 [============================>.] - ETA: 0s - loss: 1.1425e-07 - acc: 1.0000
10000/10000 [==============================] - 7s 701us/step - loss: 1.1425e-07 - acc: 1.0000 - val_loss: 1.1513e-07 - val_acc: 1.0000
25/4999 [..............................] - ETA: 0s
525/4999 [==>...........................] - ETA: 0s
875/4999 [====>.........................] - ETA: 0s
1475/4999 [=======>......................] - ETA: 0s
2100/4999 [===========>..................] - ETA: 0s
2700/4999 [===============>..............] - ETA: 0s
3225/4999 [==================>...........] - ETA: 0s
4100/4999 [=======================>......] - ETA: 0s
4675/4999 [===========================>..] - ETA: 0s
4999/4999 [==============================] - 0s 96us/step
Test score: [1.1512877142241595e-07, 1.0]
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
dense_13 (Dense) (None, 64) 704
_________________________________________________________________
dense_14 (Dense) (None, 64) 4160
_________________________________________________________________
dense_15 (Dense) (None, 1) 65
=================================================================
Total params: 4,929
Trainable params: 4,929
Non-trainable params: 0
我正在做二进制分类。以下是数据的显示方式x_train
[[0.69065374 0. 0.27677792 ... 0. 1.0274839 0.48911577]
[0.4631601 0.04829948 0.11175615 ... 0.09347356 1.8268523 0. ]
[0.7308857 0. 0.3192799 ... 0. 2.8403711 0.14755964]
...
[1.3862612 0. 1.0800421 ... 0.8344357 0.8264028 0. ]
[2.4669604 0. 1.210294 ... 1.9650785 1.3511596 0. ]
[2.246204 0. 1.1608332 ... 1.9253167 1.2738075 0. ]]
y_train：
[0. 0. 0. ... 1. 1. 1.]
首先放置等级为0的样本，然后放入等级为1的样本，因为我分别对这些样本进行了一些其他操作，然后将它们连接在一起以适合模型。
我输入数据后是否有确切的说明？