Making centralized (graph) computation faster, distributed and (at times) better thumbnail
Pause
Mute
Subtitles
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

Making centralized (graph) computation faster, distributed and (at times) better

Published on Oct 16, 20122584 Views

I will introduce a generic method for approximate inference in graphical models using graph partitioning. The resulting algorithm is linear time and provides an excellent approximation for the maximum

Related categories

Chapter list

Making Centralized (Graph) Computation Faster, Distributed And (At Times) Better00:00
Graphical models01:05
Example - 102:11
Example - 204:03
Probabilistic model05:29
Inference: Clustering - 106:35
Inference: Clustering - 209:43
Inference: Clustering - 313:40
Inference: pair-wise graphical model - 114:47
Inference: pair-wise graphical model - 215:12
Inference: pair-wise graphical model - 315:48
Inference: pair-wise graphical model - 417:40
Optimization: "prominent" methods18:38
Graphical model: "prominent" methods18:55
This talk19:34
Algorithm - 122:45
Algorithm - 223:43
Algorithm - 330:46
Graph with poly-growth31:41
Algorithm: general graph36:56
Randomized decomposition38:52
Algorithm for modularity opt41:16
Is it an excellent result?41:46
Algorithm: a useful variation - 142:53
Algorithm: a useful variation - 243:43
Example - 343:59
Example: Blondel et al44:20
Example: Blondel et al + partition44:48
Algorithm: variations - 145:24
Algorithm: variations - 247:33
Algorithm: variations - 348:10
Summary48:35