Back Propagation

Backpropagation is a supervised learning technique used for training artificial neural networks. It was first described by Paul Werbos in 1974, and further developed by David E. Rumelhart, Geoffrey E. Hinton and Ronald J. Williams in 1986.It is most useful for feed-forward networks (networks that have no feedback, or simply, that have no connections that loop). The term is an abbreviation for "backwards propagation of errors". Backpropagation requires that the transfer function used by the artificial neurons (or "nodes") be differentiable.
See more at Wikipedia.org...

(Or "backpropagation") A learning algorithm for modifying a feed-forward neural network which minimises a continuous "error function" or "objective function." Back-propagation is a "gradient descent" method of training in that it uses gradient information to modify the network weights to decrease the value of the error function on subsequent tests of the inputs. Other gradient-based methods from numerical analysis can be used to train networks more efficiently.
Back-propagation makes use of a mathematical trick when the network is simulated on a digital computer, yielding in just two traversals of the network (once forward, and once back) both the difference between the desired and actual output, and the derivatives of this difference with respect to the connection weights.

A training algorithm for multilayer perceptrons. Reliable and well-known, although significantly slower than some of the more modern algorithms (see Patterson, 1996; Fausett, 1994; Haykin, 1994).

Presenting training cases in a random order on each epoch , to prevent various undesirable effects which can otherwise occur (such as oscillation and convergence to local minima). See, Neural Networks .

back propagation
(Tech) truyền lan ngược [NB]
(C) 2007 www.TừĐiểnTiếngViệt.net