Matrix. How to correctly submit data for neural network training?

Question

Python

There is data for training the neural network. For a person, these are sets of line graphs of 4 pieces. Those. In Excel, this data can be drawn as a 100x4 matrix. How to properly submit this data for neural network training? Are there any ways to submit data in the form of a matrix, or can you simply combine them into a 400x1 vector and the network itself will figure out what's what?

Answer 1

For keras, when training LSTM, RNN or convolutional networks, such data must be supplied in the form of a tensor, i.e. make the matrix three-dimensional or larger. Any two-dimensional matrix M(X, Y) can be represented as a three-dimensional matrix M(X, Y, Z), where Z=1.
This is done using reshape - the array dimension changes when the data is saved.
M =np.reshape(M, (X, Y, Z)) or M =np.reshape(M, (X, Z, Y)
) M(X, 400), and then after loading the data we do M =np.reshape(M, (X, 4, 100)).
The result is X sets of two-dimensional 4x100 matrices.
In principle, to load into python, you can save all the data in one line to a file, and then use reshape to cut them into the desired format (2d, 3d, 4d, etc.).

Answer 2

Depends on what your object is. Judging by your description, each line is 4 different parameters of one object. And by no means can they be shoved into one vector.
And in what form to submit, this is a very vague concept. First, define a neural network. With the help of which libraries you will work with it, and from there the form for submitting information will come up by itself.