Skip to main content
Cross Validated

You are not logged in. Your edit will be placed in a queue until it is peer reviewed.

We welcome edits that make the post easier to understand and more valuable for readers. Because community members review edits, please try to make the post substantially better than how you found it, for example, by fixing grammar or adding additional resources and hyperlinks.

4
  • 1
    $\begingroup$ It works because with no hidden layer, the model parameters are identified. It’s no different than multi-class logistic regression. en.m.wikipedia.org/wiki/Identifiability $\endgroup$ Commented Mar 25 at 19:53
  • $\begingroup$ More details: stats.stackexchange.com/questions/349418/… $\endgroup$ Commented Mar 25 at 20:08
  • $\begingroup$ I'm not sure that I understand your comment. How does Identifiability explains zero initialization is better than other types of random initialization? $\endgroup$ Commented Mar 26 at 7:29
  • 1
    $\begingroup$ Each weight gets a unique update during optimization. This update moves the weights towards the global minimum. $\endgroup$ Commented Mar 26 at 11:28