I'm looking for a good review paper or book chapter that offers an accessible introduction to the computational complexity of training neural networks for classification problems. Some time back, I found a text book that stated that training an MLP network is NP-Complete, and there is this paper - but I haven't found much beyond that.
In particular, I'm trying to study questions like:
- How is training complexity related to network topology ?
- How is training complexity related to the complexity of the decision boundary?