Menu
Home
People
Places
Arts
History
Plants & Animals
Science
Life & Culture
Technology
Reference.org
Wake-sleep algorithm
open-in-new
<p>The wake-sleep algorithm is an <a href="/facts/Unsupervised_learning/8a6f61MC">unsupervised learning</a> algorithm for deep <a href="/facts/Generative_model/JUgExNIP">generative models</a>, especially <a href="/facts/Helmholtz_machine/E5G0Wzpg">Helmholtz Machines</a>. The algorithm is similar to the <a href="/facts/Expectation%E2%80%93maximization_algorithm/32P2bdBc">expectation-maximization algorithm</a>, and optimizes the model <a href="/facts/Likelihood_function/1L8P4NBX">likelihood</a> for observed data. The name of the algorithm derives from its use of two learning phases, the “wake” phase and the “sleep” phase, which are performed alternately. It can be conceived as a model for learning in the brain, but is also being applied for <a href="/facts/Machine_learning/e0w0XJTu">machine learning</a>. </p>