I am trying to understand what neural nets have in common with multiple linear regressions. Is is correct that every node in every layer (not counting the input layer) is just a multiple regression with the variables being the nodes in the preceding layer? I've included a picture to show what I mean. How is this understanding wrong/right?
