In a slide about multi-task learning, it’s said that there may be some multi-task output labels that include “?” in some features.
Why could this happen? Is it kind of null value? And how will neuron network deal with this “?” value?
Thank you.
? stands for missing label. This can happen when your labeler doesn’t provide the values for these labels.
Each column is a sample and each row is a label.
The NN would consider only consider labels with values 0 and 1 when calculating the loss.
In your case, the NN will consider 1, 0, and 1 when calculating the loss.
See https://www.coursera.org/learn/machine-learning-projects/lecture/l9zia/multi-task-learning from 4:48