john cunningham and david knowles machine learning rcc 08...

64
John Cunningham and David Knowles Machine Learning RCC 08 December 2011 Approximate Inference

Upload: others

Post on 08-Aug-2020

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

John Cunninghamand

David Knowles

Machine Learning RCC08 December 2011

Approximate Inference

Page 2: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

• Motivation

• Taxonomy

• Summations

• Estimators

• Easier Integrals

• Summary

Outline

Page 3: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

• Motivation

• Taxonomy

• Summations

• Estimators

• Easier Integrals

• Summary

Outline

Page 4: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Probabilistic Inference

• Bayes Rule:

Page 5: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Probabilistic Inference

• Bayes Rule:

• (frequentist/statistical inference)

• (Bayesian/non-Bayesian distinction)

• (conjugate models)

• (enumerable simple cases)

Page 6: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

• Many (most?) problems of interest in inference can be written as an integral of type:

• Examples:

• Posterior mean and moments:

• Data likelihood and model selection:

• Prediction:

Just an Integral

Page 7: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Central Object of Interest

Page 8: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Central Object of Focus

• Why not...

Page 9: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Central Object of Focus

• Why not...

• message passing on the factor graph?• explains {BP,VB,EP,Gibbs,etc.} nicely• abstracts approximate inference to message calculation• mechanistic, not actually the problem we are trying to solve

Page 10: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Central Object of Focus

• Why not...

• message passing on the factor graph?• explains {BP,VB,EP,Gibbs,etc.} nicely• abstracts approximate inference to message calculation• mechanistic, not actually the problem we are trying to solve

• the posterior?• pretty much the same thing, but again not often the core problem

Page 11: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

• A huge field• Bishop PRML: ~100 pages• MacKay Info Theory, Inference, ... : ~180 pages• Murphy ML: A Probabilistic Perspective: ~110 pages• MLSS: ~half a day

Fool’s Errand

Page 12: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

• A huge field• Bishop PRML: ~100 pages• MacKay Info Theory, Inference, ... : ~180 pages• Murphy ML: A Probabilistic Perspective: ~110 pages• MLSS: ~half a day

• Scope of this talk: • tutorial view of the field• incorporate by reference where possible• details where (hopefully) valuable

Fool’s Errand

Page 13: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

• Motivation

• Taxonomy

• Summations

• Estimators

• Easier Integrals

• Summary

Outline

Page 14: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Approximate Inference Taxonomy

Page 15: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Approximate Inference Taxonomy

Page 16: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Approximate Inference Taxonomy

• “Replace hard integrals with summations”

• Sampling methods• Central problem: how to

sample • Monte Carlo, MCMC,

Gibbs, etc.

Page 17: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Approximate Inference Taxonomy

• “Replace hard integrals with summations”

• Sampling methods• Central problem: how to

sample • Monte Carlo, MCMC,

Gibbs, etc.

Page 18: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Approximate Inference Taxonomy

• “Replace hard integrals with easier integrals”

• Message Passing on Factor Graph

• Central problem: how to find

• VB, EP, etc.• Note (cheat): Also

“replace hard sums with easier sums”: BP, LBP,etc.

• “Replace hard integrals with summations”

• Sampling methods• Central problem: how to

sample • Monte Carlo, MCMC,

Gibbs, etc.

Page 19: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Approximate Inference Taxonomy

• “Replace hard integrals with easier integrals”

• Message Passing on Factor Graph

• Central problem: how to find

• VB, EP, etc.• Note (cheat): Also

“replace hard sums with easier sums”: BP, LBP,etc.

• “Replace hard integrals with summations”

• Sampling methods• Central problem: how to

sample • Monte Carlo, MCMC,

Gibbs, etc.

• “Replace hard integrals with estimators”

• “Non-Bayesian” methods• Central problem: how to

find • MAP, ML, Laplace, Nested

Laplace, etc.

Page 20: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Approximate Inference Taxonomy

• “Replace hard integrals with easier integrals”

• Message Passing on Factor Graph

• Central problem: how to find

• VB, EP, etc.• Note (cheat): Also

“replace hard sums with easier sums”: BP, LBP,etc.

• “Replace hard integrals with summations”

• Sampling methods• Central problem: how to

sample • Monte Carlo, MCMC,

Gibbs, etc.

• “Replace hard integrals with estimators”

• “Non-Bayesian” methods• Central problem: how to

find • MAP, ML, Laplace, Nested

Laplace, etc.

Deterministic MethodsRandom Methods

Page 21: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

• Motivation

• Taxonomy

• Summations

• Estimators

• Easier Integrals

• Summary

Outline

Page 22: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Approximate Inference Taxonomy

• “Replace hard integrals with summations”

• Sampling methods• Central problem: how to

sample • Monte Carlo, MCMC,

Gibbs, etc.

Page 23: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Summations

• Two basic types: Sampling and MCMC

• “Instead of choosing [points] randomly, then weighting them..., we choose [points] with a probability... and weight them evenly.” - Metropolis et al (1953).

Page 24: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Summations

• Sampling:

Page 25: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Summations

• Sampling:

Page 26: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Summations

• Sampling:

• “pick an arbitrary point and weight it by what you care about.”

• MC, importance, rejection.

Page 27: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

• MH/MCMC:

Summations

• Sampling:

• “pick an arbitrary point and weight it by what you care about.”

• MC, importance, rejection.

Page 28: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

• MH/MCMC:

Summations

• Sampling:

• “pick an arbitrary point and weight it by what you care about.”

• MC, importance, rejection.

Page 29: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

• MH/MCMC:

Summations

• Sampling:

• “pick an arbitrary point and weight it by what you care about.”

• MC, importance, rejection.

Page 30: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

• MH/MCMC:

• “pick a point from what you care about and weight it evenly.”

• MH, MCMC, AIS, Gibbs, HMC, Slice Sampling, ESS, Hamiltonian MCMC, RML, ...

Summations

• Sampling:

• “pick an arbitrary point and weight it by what you care about.”

• MC, importance, rejection.

Page 31: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Big Topic, Incorporated by Reference

• Iain Murray’s MLSS lectures: http://videolectures.net/mlss09uk_murray_mcmc/

Page 32: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

• Motivation

• Taxonomy

• Summations

• Estimators

• Easier Integrals

• Summary

Outline

Page 33: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Approximate Inference Taxonomy

• “Replace hard integrals with estimators”

• “Non-Bayesian” methods• Central problem: how to

find • MAP, ML, Laplace, Nested

Laplace, etc.

Page 34: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Approximate Inference Taxonomy

• “Replace hard integrals with estimators”

• “Non-Bayesian” methods• Central problem: how to

find • MAP, ML, Laplace, Nested

Laplace, etc.

• Laplace

Page 35: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Approximate Inference Taxonomy

• “Replace hard integrals with estimators”

• “Non-Bayesian” methods• Central problem: how to

find • MAP, ML, Laplace, Nested

Laplace, etc.

• Laplace

Page 36: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

• MAP

Approximate Inference Taxonomy

• “Replace hard integrals with estimators”

• “Non-Bayesian” methods• Central problem: how to

find • MAP, ML, Laplace, Nested

Laplace, etc.

Page 37: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

• Nested Laplace• Rue and Martino (2009), “Approximate

Bayesian Inference for latent Gaussian models by using integrated nested Laplace approximations”, JRSSB.

Approximate Inference Taxonomy

• “Replace hard integrals with estimators”

• “Non-Bayesian” methods• Central problem: how to

find • MAP, ML, Laplace, Nested

Laplace, etc.

Page 38: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

• Nested Laplace• Rue and Martino (2009), “Approximate

Bayesian Inference for latent Gaussian models by using integrated nested Laplace approximations”, JRSSB.

• ...but see Cseke and Heskes (2011), “Approximate marginals in latent Gaussian models”, JMLR.

Approximate Inference Taxonomy

• “Replace hard integrals with estimators”

• “Non-Bayesian” methods• Central problem: how to

find • MAP, ML, Laplace, Nested

Laplace, etc.

Page 39: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

• Motivation

• Taxonomy

• Summations

• Estimators

• Easier Integrals

• Summary

Outline

Page 40: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Approximate Inference Taxonomy

• “Replace hard integrals with easier integrals”

• Message Passing on Factor Graph

• Central problem: how to find

• VB, EP, etc.• Note (cheat): Also

“replace hard sums with easier sums”: BP, LBP,etc.

Page 41: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Message Passing on Factor Graph

Page 42: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Belief Propagation / Sum-product

Page 43: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Approximate messages

Page 44: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Expectation Propagation (EP)

Page 45: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Expectation Propagation (EP)

Page 46: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Instead...

~

~

Page 47: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Instead, EP does this...

Moment Match

A - Form “CAVITY” B - Add a true factor and “PROJECT”

~

~

Page 48: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Instead, EP does this...

Moment Match

A - Form “CAVITY” B - Add a true factor and “PROJECT”

Page 49: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

At convergence, we have...

~

~

Page 50: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Approximately this...

Moment Match

Page 51: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Beyond Simple EP

Page 52: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Variational Bayes / Variational Message Passing

Page 53: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Variational Bayes / Variational Message Passing

Page 54: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Summary of Message Passing Perspective

Page 55: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

• Exclusive (VB mode-seeking) vs. inclusive (EP) KL, consequences for multimodality

• Damping for EP• Power EP• More structured approximations (GBP, tree EP,

structured VB)• Connection to EM • Infer.NET

Things to be aware of

Page 56: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

• Motivation

• Taxonomy

• Summations

• Estimators

• Easier Integrals

• Summary

Outline

Page 57: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Approximate Inference Taxonomy

• “Replace hard integrals with easier integrals”

• Message Passing on Factor Graph

• Central problem: how to find

• VB, EP, etc.• Note (cheat): Also

“replace hard sums with easier sums”: BP, LBP,etc.

• “Replace hard integrals with summations”

• Sampling methods• Central problem: how to

sample • Monte Carlo, MCMC,

Gibbs, etc.

• “Replace hard integrals with estimators”

• “Non-Bayesian” methods• Central problem: how to

find • MAP, ML, Laplace, Nested

Laplace, etc.

Page 58: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Summary of Features

Page 59: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Summary of Features

• exact (eventually)

• fast/efficient in big-huge cases (at times the only option)

• poor for model selection

• slow error convergence

Page 60: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

• analytically useful

• fits into many MLschemes (bounds)

• fast/efficient in small-medium cases

• no exactness (ignores some features of true integral)

Summary of Features

• exact (eventually)

• fast/efficient in big-huge cases (at times the only option)

• poor for model selection

• slow error convergence

Page 61: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

• analytically useful

• fits into many MLschemes (bounds)

• fast/efficient in small-medium cases

• no exactness (ignores some features of true integral)

Summary of Features

• exact (eventually)

• fast/efficient in big-huge cases (at times the only option)

• poor for model selection

• slow error convergence

• quick and dirty

• often works well

• quick and dirty (local... ignores many features of true integral)

Page 62: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

• Has many names and duplicate fields, but in the end is just numerical integration

• Disappointingly (necessarily?) fractured field

• Inherently problem-specific

Conclusion

Page 63: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning

Resources• Books

• Bishop (2006) “Pattern Recognition and Machine Learning”, Chapters 10-11• Murphy (2012) “ML: A Probabilistic Perspective”, Chapters 18 - 22

• Rasmussen and Williams (2006), “Gaussian Processes for Machine Learning”, Chapter 3 (for EP and Laplace).

• MacKay (2003) “Information Theory, Inference, and Learning Algorithms”, Part IV

• Video• MLSS 09: Murray (MCMC): http://videolectures.net/mlss09uk_murray_mcmc/

• MLSS 09: Minka (Min Divergence): http://videolectures.net/mlss09uk_minka_ai/

• Papers• Wainwright and Jordan (2008), “Graphical Models, Exponential Families, and

Variational Inference.” Foundations and Trends in Machine Learning.

• Winn and Bishop (2005), “Variational Message Passing”, JMLR.• Minka and Winn (2009): Gates, NIPS

• Hennig (2011), “Approximate Inference in Graphical Models” (PhD Thesis), Chap 2.• Kuss and Rasmussen (2005), “Assessing Approximate Inference for Binary

Gaussian Process Classification”, JMLR.

Page 64: John Cunningham and David Knowles Machine Learning RCC 08 ...cbl.eng.cam.ac.uk/pub/Intranet/MLG/ReadingGroup/20111208cambri… · John Cunningham and David Knowles Machine Learning