In the video 16.2, Hinton gives a really high level explanation of why we need to halve the weights of some layers when we combine RBMs to build a DBM. Can you give a more detailed explanation of why a simple division by 2 is a suitable geometric mean?
Posts Tagged 'DBM'
DBM and doubled weights
Published April 21, 2013 apr22 , quiz 1 CommentTags: DBM, Xavier Bouthillier
Representation Learning 