2 views (last 30 days)
I want to train and test a 1D convolutional network for sequence - to - sequence classification.
I have the following architecture:
layers = [ ...
If I include the globalAveragePooling1dLayer after my second relu layer than i get the following error:
" Error using trainNetwork (line 184) Invalid training data. For image, sequence-to-label, and feature classification tasks, responses must be categorical" .
Without the globalAveragePooling1dLayer I don't get the error and trainings starts. What is the problem?
Tomaso Cetto on 18 Jan 2022
Edited: Tomaso Cetto on 18 Jan 2022
As you've noticed, the globalAveragePooling1dLayer plays a critical role here. This is because that layer removes the time dimension by pooling over it globally (i.e. keeping only the largest value in the sequence). This layer is useful for sequence-to-one tasks, where the output isn't a sequence. The output here would be a numClasses x numObservations array.
However, because yours is a sequence-to-sequence problem, you want the output to be a numClasses x numObservations x sequenceLength array (with the sequence dimension conserved). So in that case, the globalAveragePooling1dLayer isn't appropriate for your workflow, because of the fact it removes the sequence dimension.
Hope this helps, and let me know if you have any other questions!