en
0.25
0.5
0.75
1.25
1.5
1.75
2
Restricted Deep Belief Networks for Multi-View Learning
Published on Oct 03, 20113248 Views
Deep belief network (DBN) is a probabilistic generative model with multiple layers of hidden nodes and a layer of visible nodes, where parameterizations between layers obey harmonium or restricted B
Related categories
Chapter list
Restricted Deep Belief Network for Multi-view Learning00:00
Multi-view Data00:19
Multi-view Learning01:30
Our Approach02:32
Exponential Family Harmonium (EFH)03:38
Dual-wing Harmonium (DWH)05:07
Proposed Model: Multi-view Harmonium06:03
Training Multi-view Harmonium (1)07:04
Training Multi-view Harmonium (2)08:26
Restricted Deep Belief Network08:58
Inference on RDBN10:35
Inferring Latent Variables from Synthetic Data (1)11:45
Inferring Latent Variables from Synthetic Data (2)12:55
Image Reconstruction on NORB-small Dataset (1)13:30
Image Reconstruction on NORB-small Dataset (2)14:25
Image Annotation on ESL Dataset (1)15:06
Image Annotation on ESL Dataset (2)16:03
Conclusions and Future Work16:30