en-de
en-es
en-fr
en-sl
en
en-zh
0.25
0.5
0.75
1.25
1.5
1.75
2
Spectral Dimensionality Reduction via Maximum Entropy, incl. discussion by Laurens van der Maaten
Published on May 06, 20115417 Views
We introduce a new perspective on spectral dimensionality reduction which views these methods as Gaussian random fields (GRFs). Our unifying perspective is based on the maximum entropy principle whi
Related categories
Chapter list
Spectral Dimensionality Reduction via Maximum Entropy00:00
Outline00:35
Maximum Entropy Unfolding00:38
Spectral Approaches (1)00:39
Spectral Approaches (2)01:06
Spectral Approaches (3)01:35
Spectral Approaches (4)01:52
Spectral Approaches (5)01:54
Classical MDS and KPCA (1)02:15
Classical MDS and KPCA (2)02:40
Classical MDS and KPCA (3)02:55
Maximum Variance Unfolding (1)03:41
Maximum Variance Unfolding (2)03:50
Maximum Variance Unfolding (2)03:51
Maximum Variance Unfolding (3)03:52
Maximum Variance Unfolding (4)03:58
Maximum Entropy Unfolding (1)04:35
Maximum Entropy Unfolding (2)04:49
Maximum Entropy Unfolding (3)05:00
Maximum Entropy Unfolding (4)05:02
Maximum Entropy Unfolding (5)05:34
Maximum Entropy Unfolding (6)06:16
Gaussian Random Field (1)06:51
Gaussian Random Field (2)07:21
Gaussian Random Field (3)07:37
Gaussian Random Field (4)07:54
Gaussian Random Field (5)08:01
Data Feature Independence08:20
Blessing of Dimensionality (1)09:11
Blessing of Dimensionality (2)09:15
Blessing of Dimensionality (3)09:16
Blessing of Dimensionality (4)09:20
Blessing of Dimensionality (4)09:22
Blessing of Dimensionality (5)09:28
Blessing of Dimensionality (6)09:52
Relations to Other Spectral Methods10:16
Relationship to Laplacian Eigenmaps (1)10:18
Relationship to Laplacian Eigenmaps (2)10:19
Relationship to Laplacian Eigenmaps (3)10:20
Relationship to Laplacian Eigenmaps (4)10:20
Relationship to Laplacian Eigenmaps (5)10:21
Relationship to Laplacian Eigenmaps (6)10:22
Locally Linear Embedding (1)11:19
Locally Linear Embedding (2)11:23
Locally Linear Embedding (3)11:30
Locally Linear Embedding (4)11:43
Locally Linear Embedding (5)11:45
Locally Linear Embedding (6)12:10
Locally Linear Embedding (7)12:16
Locally Linear Embedding (8)12:25
LLE Approximates MEU (1)12:37
LLE Approximates MEU (2)12:48
LLE Approximates MEU (3)12:50
LLE Approximates MEU (4)12:54
LLE Approximates MEU (5)13:00
LLE Approximates MEU (6)13:05
LLE and PC (1)13:12
LLE and PC (2)13:15
LLE and PC (3)13:17
LLE and PC (4)13:19
Isomap (1)13:54
Isomap (2)13:58
Isomap (3)14:00
Isomap (4)14:01
Isomap (5)14:02
Relation to GP-LVM (1)14:34
Relation to GP-LVM (2)14:35
Relation to GP-LVM (3)14:36
Relation to GP-LVM (4)14:37
Experiments14:38
Simple Experiments14:41
Motion Capture Data14:44
Laplacian Eigenmaps and LLE15:06
Isomap and MVU15:15
MEU15:20
Motion Capture: Model Scores15:24
Robot Navigation Example15:52
Laplacian Eigenmaps and LLE16:17
Isomap and MVU16:19
MEU16:25
Robot Navigation: Model Scores16:48
Discussion and Conclusions16:59
Stages of Spectral Dimensionality Reduction (1)17:04
Stages of Spectral Dimensionality Reduction (2)17:11
Stages of Spectral Dimensionality Reduction (3)17:26
Stages of Spectral Dimensionality Reduction (4)17:45
Our Perspective (1)18:34
Our Perspective (2)18:35
Our Perspective (3)18:36
Our Perspective (4)18:37
Our Perspective (5)18:39
Advantages of Existing Approaches (1)18:42
Advantages of Existing Approaches (2)18:45
Advantages of Existing Approaches (3)18:53
Acknowledgements19:14
Discussion of “Spectral Dimensionality Reduction via Maximum Entropy”19:30
Timeline19:43
Manifold learning vs. generative modeling (1)21:24
Manifold learning vs. generative modeling (2)22:42
Manifold learning vs. generative modeling (3)23:38
Rank minimization (1)25:02
Rank minimization (2)26:01