Concatinate 4D Time-Signals

Hi,

How can i concatinate 20 , 4D signals

I1 = rand(32,32,1,13) ;
I2 = rand(40,32,1,13) ;
...
I20 = rand(47,32,1,13) ;
Catq1=cat(4,I1,I2);

The above give me error? What can i do?

Answer 1

[Ridwan Alam]

You want to concatenate on the dim which is not same for both, right?

Catq1 = cat(1,I1,I2);

% OR

Catq1 = [I1;I2];

[ARN]

Can you tell me if i am on the right track? So i am extracting data from lets video and i get 3-D data (512 points, 4 channels, 320 timeseries) for 200+ cycles and likewise i am getting for next 20 files.

what i am trying to do is concatinate all the timeseries which will become 320*200*20 and each time-sample i have 520 points on all 4 channels.

ldm= length(data);
[Ns, Nch, Rs] = size(data(1).FL.TimeSignals);
a=zeros(ldm,Ns,Nch,Rs); 
for i=1: ldm
    a(i,:,:,:)=data(i).FL.TimeSignals, Ns, Nch, Rs;
end
I1=a(:,:,1,:);

a size will be (200, 512,4, 320 ). Now same procedure for all 20 videos and then concatinate it like you proposed?

[Ridwan Alam]

well .. without looking at the 'data', it's a bit of a guess .. but looks fine to me, except a few questions. the rehsape() inside the loop is not really doing anything, right? because all data(i).FL.TimeSignals have size [Ns, Nch, Rs]. Now, what are you trying to do with this line: I1=a(:,:,1,:); this will make I1 of size [ldm,Ns,Rs].

[ARN]

No, reshape is not there.. i wrote it wrong. Thanks for the reply

[Ridwan Alam]

Sure. Glad to help!

Answer 2

[Bartlomiej Mroczek]

And how to provide the CNN: predict DL block to the ready machine.

1. I have data built in 4D format, my CNN machine only use this format for learning and prediction in matlab.

2. I have built a model in SIMULINK but there is a problem with entering data to the Predict block, error message:

Available formats are double non-complex matrix, a structure with or without time, or a structure with MATLAB timeseries as leaf nodes. All formats require the data to be finite (not Inf or NaN).

How to solve it?