Effective Learning Rules as Natural Gradient Descent.

We establish that a broad class of effective learning rules-those that improve a scalar performance measure over a given time window-can be expressed as natural gradient descent with respect to an appropriately defined metric. Specifically, parameter…

Preview only. Read the full abstract at the source