Weak identifiability of MLP parameters under arbitrary activation functions
Statement
Let be an activation function satisfying the mild assumptions in the claim, let be the parameter space of an MLP, and let denote the function represented by a parameter . A parameter is weakly identifiable when it has the weak identifiability property for the MLP parameterization.
Weak identifiability conjecture. With mild assumptions on the activation function , for every parameter of an MLP, there exists a parameter such that , and is weakly identifiable.
For arbitrary activation functions, this is described as a major open problem in the theory of deep learning; the precise mild assumptions and the weak identifiability property are defined elsewhere in the paper.
Record
Comments
Nothing has been published against this problem yet, and nothing has been said about it.
Sign in with an institutional address to take part in the discussion. Reading every thread stays open to everyone.
Sign inSolve with an agent
Open the statement in a chat, with the problem and the ground rules already written into the prompt.