Link to slides: docs.google.com/presentation/d/1jCY9-_vGkUV1wFcHxp07lWNF6XMITMZIiYdYnYT6IHs/edit?usp=sharing&resourcekey=0-WceWVLKaJMiJ0VLXoPXANw
@thomashm9990
Жыл бұрын
I am still confused about the use of those z indexes. To what extent are they helping the network to know what it doesn't know?
@mgostIH
Жыл бұрын
You can think of the z index as selecting a different model from a population each with a different prior belief. By observing where they all agree, you know that part has had enough data to have them converge, if somewhere else the guesses for each model in the population are wildly different it means the data wasn't enough to overcome the individuals' prior beliefs.
Пікірлер: 4