Concatenate- x1은 1에서 100, 311에서 410, 1에서 100의 총 (300, 3)의 모양
- x2는 101에서 200, 311에서 410, 101에서 200의 총 (300, 3)의 모양
- 이 두 데이터셋이 앙상블이 되어 501에서 600이 나오는 구조
- 각각 300개씩의 데이터를 가지고 있는 2개의 x
- 100개의 데이터를 가지고 있는 1개의 y
train, test, validation 분리Model: "model"
__________________________________________________________________________________________________
Layer (type) Output Shape Param # Connected to
==================================================================================================
input_1 (InputLayer) [(None, 3)] 0
__________________________________________________________________________________________________
dense (Dense) (None, 100) 400 input_1[0][0]
__________________________________________________________________________________________________
input_2 (InputLayer) [(None, 3)] 0
__________________________________________________________________________________________________
dense_1 (Dense) (None, 30) 3030 dense[0][0]
__________________________________________________________________________________________________
dense_3 (Dense) (None, 50) 200 input_2[0][0]
__________________________________________________________________________________________________
dense_2 (Dense) (None, 7) 217 dense_1[0][0]
__________________________________________________________________________________________________
dense_4 (Dense) (None, 7) 357 dense_3[0][0]
__________________________________________________________________________________________________
concatenate (Concatenate) (None, 14) 0 dense_2[0][0]
dense_4[0][0]
__________________________________________________________________________________________________
dense_5 (Dense) (None, 10) 150 concatenate[0][0]
__________________________________________________________________________________________________
dense_6 (Dense) (None, 5) 55 dense_5[0][0]
__________________________________________________________________________________________________
dense_7 (Dense) (None, 1) 6 dense_6[0][0]
==================================================================================================
Total params: 4,415
Trainable params: 4,415
Non-trainable params: 0
__________________________________________________________________________________________________
Train on 60 samples, validate on 20 samples
Epoch 1/100
60/60 [==============================] - 1s 10ms/sample - loss: 62526.2501 - mse: 62526.2383 - val_loss: 612.0000 - val_mse: 611.9999
Epoch 2/100
60/60 [==============================] - 0s 2ms/sample - loss: 499.4471 - mse: 499.4472 - val_loss: 1948.4065 - val_mse: 1948.4062
...
60/60 [==============================] - 0s 2ms/sample - loss: 8.9965 - mse: 8.9965 - val_loss: 3.0603 - val_mse: 3.0603
Epoch 100/100
60/60 [==============================] - 0s 2ms/sample - loss: 38.3114 - mse: 38.3114 - val_loss: 4.3581 - val_mse: 4.3581
20/20 [==============================] - 0s 1ms/sample - loss: 17.3456 - mse: 17.3456
loss : 17.345600885152816
mse : 17.3456
[581] [582.05133]
[582] [583.33923]
[583] [584.6271]
[584] [585.91486]
[585] [587.2028]
[586] [588.49054]
[587] [589.77844]
[588] [591.06635]
[589] [592.3542]
[590] [593.64197]
[591] [594.9298]
[592] [596.21765]
[593] [597.50555]
[594] [598.79346]
[595] [600.0893]
[596] [601.40186]
[597] [602.7145]
[598] [604.02704]
[599] [605.3395]
[600] [606.65216]
|