| Revision as of 05:01, 22 November 2012 edit14.139.122.114 (talk)No edit summary← Previous edit | Revision as of 20:45, 12 June 2013 edit undoQwertyus (talk | contribs)Extended confirmed users31,640 edits see also RBMs, the training algorithm for which (CD) was inspired by wake-sleepNext edit → | ||
| Line 1: | Line 1: | ||
| The '''wake-sleep algorithm''' is an ] algorithm for a multilayer ]. Training is divided into two phases, "wake" and "sleep". In the "wake" phase, neurons are driven by recognition connections (connections from what would normally be considered an input to what is normally considered an output), while generative connections (those from outputs to inputs) are modified to increase the probability that they would reconstruct the correct activity in the layer below (closer to the sensory input). In the "sleep" phase the process is reversed: neurons are driven by generative connections, while recognition connections are modified to increase the probability that they would produce the correct activity in the layer above (further from sensory input). | The '''wake-sleep algorithm''' is an ] algorithm for a multilayer ]. Training is divided into two phases, "wake" and "sleep". In the "wake" phase, neurons are driven by recognition connections (connections from what would normally be considered an input to what is normally considered an output), while generative connections (those from outputs to inputs) are modified to increase the probability that they would reconstruct the correct activity in the layer below (closer to the sensory input). In the "sleep" phase the process is reversed: neurons are driven by generative connections, while recognition connections are modified to increase the probability that they would produce the correct activity in the layer above (further from sensory input). | ||
| ==See also== | |||
| * ], a type of neural net that is trained with a conceptually similar algorithm]] | |||
| ==External links== | ==External links== | ||
Revision as of 20:45, 12 June 2013
The wake-sleep algorithm is an unsupervised learning algorithm for a multilayer neural network. Training is divided into two phases, "wake" and "sleep". In the "wake" phase, neurons are driven by recognition connections (connections from what would normally be considered an input to what is normally considered an output), while generative connections (those from outputs to inputs) are modified to increase the probability that they would reconstruct the correct activity in the layer below (closer to the sensory input). In the "sleep" phase the process is reversed: neurons are driven by generative connections, while recognition connections are modified to increase the probability that they would produce the correct activity in the layer above (further from sensory input).
See also
- Restricted Boltzmann machine, a type of neural net that is trained with a conceptually similar algorithm]]
External links
 | This artificial intelligence-related article is a stub. You can help Misplaced Pages by expanding it. |