This is machine translation

Translated by Microsoft
Mouse over text to see original. Click the button below to return to the English verison of the page.


Concatenate neural network data samples


[x1 x2 ... xn]


catsamples(x1,x2,...,xn) takes any number of neural network data values, and merges them along the samples dimension (i.e., the matrix column dimension).

If all arguments are matrices, this operation is the same as [x1 x2 ... xn].

If any argument is a cell array, then all non-cell array arguments are enclosed in cell arrays, and then the matrices in the same positions in each argument are concatenated.

catsamples(x1,x2,...,xn,'pad',v) allows samples with varying numbers of timesteps (columns of cell arrays) to be concatenated by padding the shorter time series with the value v, until they are the same length as the longest series. If v is not specified, then the value NaN is used, which is often used to represent unknown or don't-care inputs or targets.


This code concatenates the samples of two matrix data values.

x1 = [1 2 3; 4 7 4]
x2 = [5 8 2; 4 7 6]
y = catsamples(x1,x2)

This code concatenates the samples of two cell array data values.

x1 = {[1:3; 4:6] [7:9; 10:12]; [13:15] [16:18]}
x2 = {[2 1 3; 5 4 1] [4 5 6; 9 4 8]; [2 5 4] [9 7 5]}
y = catsamples(x1,x2)

Here the samples of two cell array data values, with unequal numbers of timesteps, are concatenated.

x1 = {1 2 3 4 5};
x2 = {10 11 12};
y = catsamples(x1,x2,'pad')

Introduced in R2010b

Was this topic helpful?