What does observationDim mean?

Dear All!

I am trying to train a deep learning network in the deep network designer but I got the error: "Trainig with trainNetwork failed. observationDim should be either 2, 4 or 5."

I haven't found anything about observationDim in the documentation. How can I solve this problem?

For context I am doing spine segmentation on nifti volumes.

Thank you in advance,

Daniel

[Conor Daly]

Hi Daniel,

Thanks for your message. It looks like this is an unintended error. Would you be able to share the code you ran so that we can investigate the issue further?

All the best,

Conor

[Daniel Csata]

Hi Conor,

I attach my network but I only ran the training from the deep network designer GUI, so I have no code of that. I am using a randomPatchDataStore for the training data, which is from the labels and the nifti volumes.

Thank you for looking into it!

Best regards,

Daniel

Answer 1

[Ganapathi Subramanian]

Hi Daniel,

It is my understanding that you are working on a deep learning model and wants to know about ‘observationDim’.

`observationDim` specifies the dimension of the input data that corresponds to the observations or samples in your experimental setup. For example, if your input data `X` has dimensions `numFeatures` × `numSamples`, then `observationDim` would be set to 2.