Htydjtk

now,I have a 3600*119 matrix.3600 means samples,119 means variable numbers.
I hope I can input a [1,1:119] variable, and get the next,[2,1:199],just like it.
the dataset parts:

This is totally possible with deep neural networks. You can have an input vector and an output vector. However, your specific case does not have sufficient data to train such a model. With 119 inputs and 119 outputs you will need quite a complex network architecture to capture these mappings. In general I like to have at a very minimum 100 times more instances than features when training a neural network.

Here is an example of a neural network which can sort numbers, it has 4 input neurons and 4 output neurons.

from __future__ import print_function

import keras

from keras.datasets import mnist

from keras.models import Sequential

from keras.layers import Dense, Dropout, Flatten

from keras.layers import Conv1D, MaxPooling1D, Reshape

from keras.callbacks import ModelCheckpoint

from keras.models import model_from_json

from keras import backend as K

First we will create some data

import numpy as np



n = 100000

x_train = np.zeros((n,4))

for i in range(n):

    x_train[i,:] = np.random.permutation(50)[0:4]



x_train = x_train.reshape(n, 4,)

y_train = np.sort(x_train, axis=1).reshape(n, 4,)



n = 10000

x_test = np.zeros((n,4))

for i in range(n):

    x_test[i,:] = np.random.permutation(50)[0:4]



x_test = x_test.reshape(n, 4,)

y_test = np.sort(x_test, axis=1).reshape(n, 4,)



print(x_test[0].T)

print(y_test[0])

[25. 33. 4. 2.]
[ 2. 4. 25. 33.]

Let's make our neural network model

input_shape = (4,)



model = Sequential()

model.add(Dense(32, activation='relu',

                 input_shape=input_shape))

model.add(Dense(32, activation='relu',

                 input_shape=input_shape))

model.add(Dense(32, activation='relu',

                 input_shape=input_shape))

model.add(Dense(4))



model.compile(loss=keras.losses.mean_squared_error,

              optimizer=keras.optimizers.Adadelta(),

              metrics=['accuracy'])

Now we train it

epochs = 20

batch_size = 512

# Fit the model weights.

history = model.fit(x_train, y_train,

          batch_size=batch_size,

          epochs=epochs,

          verbose=1,

          validation_data=(x_test, y_test))

You can see the progression of the training using

import matplotlib.pyplot as plt

%matplotlib inline



plt.figure(figsize=(20,5))

# summarize history for accuracy

plt.subplot(1,2,1)

plt.plot(history.history['acc'])

plt.plot(history.history['val_acc'])

plt.title('model accuracy')

plt.ylabel('accuracy')

plt.xlabel('epoch')

plt.legend(['train', 'validation'], loc='lower right')



# summarize history for loss

plt.subplot(1,2,2)

plt.plot(history.history['loss'])

plt.plot(history.history['val_loss'])

plt.title('model loss')

plt.ylabel('loss')

plt.xlabel('epoch')

plt.legend(['train', 'validation'], loc='upper right')

plt.show()

Now we have our model, we can sort new inputs

np.round(model.predict(np.array([[1,5,2,3]])))

array([[1., 2., 3., 5.]], dtype=float32)

np.round(model.predict(np.array([[20,7,24,3]])))

array([[ 3., 7., 20., 23.]], dtype=float32)