CNN WK4 One-shot learning

Course Q&A Deep Learning Specialization Convolutional Neural Networks
1

Hi I have some questions on how the following scenarios are handled for real-time one-shot learning application:

  1. The hyperparameter tau is used as a threshold for same/different person. How is the value of tau being selected usually as you want to make sure that there’s nearly impossible to produce multiple results of same person?
  2. How does one-shot learning handle identical twins scenario?
  3. Has there been any testing against full-face prosthetic mask?

An unrelated question: can we use latex formatting in this forum? It works in edX but I’m not sure about Coursera.

2
  1. You’d have to set the threshold value through model validation.
  2. Badly.
  3. Somewhere, someone has probably tried it. You’d have to do some literature searches.

Re: LaTeX; Yes, you can use it here. No, this forum platform isn’t Coursera - it’s Discourse.