A Completed Information Projection Interpretation of Expectation Propagation thumbnail
Pause
Mute
Subtitles
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

A Completed Information Projection Interpretation of Expectation Propagation

Published on Dec 31, 20074428 Views

This talk presents an interpretation of expectation propagation (EP) as a hybrid between two different iterated Bregman projections algorithms from the convex analysis and programming literature whose

Related categories

Chapter list

A Completed Information Projection Interpretation of Expectation Propagation00:00
Exponential Family Densities & Rudimentary Information Geometry00:06
Expectation Propagation01:11
Bregman Divergences02:33
Method of Alternating Bregman Projections04:32
Dykstra’s Algorithm with Cyclic Bregman Projections05:02
Two Sets Related to EP06:45
Actual Sets P & Q for 2 Bits08:32
EP as a Hybrid Algorithm - 108:41
EP as a Hybrid Algorithm - 209:53
What Does All This Mean10:21
Relationship to Prior Work (What Was Innovated)11:12
So You Don’t Go Home Hungry: Nonlinear Block Gauss Seidel Connection12:32
Convergence Theorem by Applying NLBGS Theory16:23
References17:43