Invalid validation data. For a recurrent layer with output mode 'last', responses must be a categorical column vector.

I am trying to train LSTM network following the example given in mathworks entitled "Classify Videos Using Deep Learning". But I am getting an error. I am attaching here the screenshot of the error.

In this screenshot the error is responses must be a categorical column vector. In my case, as you can see in the code (line no. 154) responses are named as labelsTrain. In the workspace you can see labelsTrain is categorical. I am not getting the reason behind this error and hence unable to find a way out of it. Please suggest some solution for this.

If my query is not clear, feel free to comment.

Answer 1

[Walter Roberson]

Your Validation Data responses that you are passing through your options is a categorical row vector, but needs to be a categorical column vector.

[NAVNEET NAYAN]

In the code line no. 154 I have already taken the transpose to make it the categorical column vector

[Walter Roberson]

You transpose the training label data in line 154. That is not the validation data. The validation data is line 149.

[NAVNEET NAYAN]

Aah... Got it. I was making a silly mistake. Thank You for your time and response.