could to create the categorical values for 2x2 cell array

I having cell array of 15x1. A (input training data for clustering)=15×1 cell array

{2×1 double}

{2×1 double}

{2×1 double}

{2×1 double}

{2×1 double}

{2×2 double}

{2×2 double}

{2×2 double}

{2×2 double}

{2×2 double}

{2×3 double}

{2×3 double}

{2×3 double}

{2×3 double}

{2×3 double}

B(targets for clustering) needs to be in the following manner.

[ 1

1

1

1

1

1 1

1 2

2 1

2 1

1 2

1 2 2

2 1 1

1 3 2

1 1 2

2 2 1]

Could you please help me to get it.

[Rik]

Feel free to reopen this question to edit it and make it more clear.

Answer 1

[Walter Roberson]

B = cellfun(@convert_to_categorical, A, 'uniform', 0);

function C = convert_to_categorical(M)
   if iscell(M)
      C = cellfun(@convert_to_categorical, M, 'uniform', 0);
   else
      C = categorical(M);
   end
end

[jaah navi]

I want to have B as follows

B(targets for clustering)

[ 1 for {2×1 double}

1 for {2×1 double}

1 for {2×1 double}

1 for {2×1 double}

1 for {2×1 double}

1 1 for {2×2 double}

1 2 for {2×2 double}

2 1 for {2×2 double}

2 1 for {2×2double}

1 2 for {2×2 double}

1 2 2 for {2×3 double}

2 1 1 for {2×3 double}

1 3 2 for {2×3double}

1 1 2 for {2×3 double}

2 2 1] for {2×3double}

Could you please help me to get it.

[Walter Roberson]

When you look at the data at B{12} compared to B{13} how can you tell whether the targets should be [2 1 1] or [1 3 2] ?

[jaah navi]

sorry I made a mistake actually my targets are

[ 1

1

1

1

1

1 1

1 2

2 1

2 1

1 2

1 2 2

2 1 1

1 3 2

1 1 2

2 2 1]

which are collected before

but I doesn't know how to train the model based on the inputs containing cell array with its corresponding targets shown above.

Could you please help me to organise it.

[Walter Roberson]

Sorry, I still have no idea how the content of B{12} leads you to know to use [2 1 1] but the content of B{13} leads you to know to use [1 3 2] as the targets.

If you (somehow) knew you had

   targ = {  1	
        1	
        1	
        1	
        1	
        [1	1]
        [1	2]	
        [2	1]	
        [2	1]	
        [1	2]	
        [1	2	2]
        [2	1	1]
        [1	3	2]
        [1	1	2]
        [2	2	1]
        }
   

and then you would

   ctarg = cellfun(@categorical, targ, 'uniform', 0);

after which you would pass ctarg in as the target data.

[jaah navi]

i used ctarg(B) command for my targets. thanks for it.

But when I implement it to train the model using the following code

A
 B
inputSize = 2;
numHiddenUnits = 100;
numClasses = 3;
layers = [ ...
sequenceInputLayer(inputSize)
fullyConnectedLayer(3)
reluLayer
regressionLayer]
maxEpochs = 50;
miniBatchSize = 50;
options = trainingOptions('adam', ...
    'ExecutionEnvironment','cpu', ...
    'MaxEpochs',maxEpochs, ...
    'MiniBatchSize',miniBatchSize, ...
    'GradientThreshold',1, ...
    'Verbose',false, ...
    'Plots','training-progress');
net = trainNetwork(XTrainCopy,ctarg,layers,options);

i am getting error stating Invalid training data. Responses must be a matrix of numeric responses, or a N-by-1 cell array of sequences, where N is the number of sequences. The feature dimension of all sequences must be the same.

Error in seven (line 21)

net = trainNetwork(A,B,layers,options);

Caused by:

Error using nnet.internal.cnn.util.NetworkDataValidator/assertValidSequenceResponse (line

356)

Invalid training data. Responses must be a matrix of numeric responses, or a N-by-1 cell

array of sequences, where N is the number of sequences. The feature dimension of all

sequences must be the same.

Could you please help me to overcome it.

[jaah navi]

If my code is wrong please guide me how to model the code to train my inputs and targets.