wanderer comments on resolving some neural network mysteries

[wanderer]

Adding enough free variables can turn non-convex problems into convex ones.

Could you explain why this is true?

[bhauth]

see this post