MachineLearning

These pages have been compiled by members of the CBU Learning Machine Learning (LML) Group

Learning Machine Learning Course

1. Introduction (applications, supervised, unsupervised, semi-supervised, reinforcement learning, bayes rule, probability theory, randomness) Presentation1_LML.ppt , 27 May 2008, Eleftherios Garyfallidis.

2. Further Introduction (what is ML, bayes rule, bayesian regression,entropy, relative entropy, mutual information), Presentation2_LML.ppt , 3 June 2008, Eleftherios Garyfallidis.

3. Maximum Likelihood vs Bayesian Learning (Notes available upon request) Presentation3_LML.ppt , 10 June 2008, Hamed Nili.

4. Factor Analysis, PCA and pPCA, Presentation4_LML.ppt , 17 June 2008, Hamed Nili.

5. Independent Component Analysis (ICA), Presentation5_LML.pdf , 24 June 2008, Jason Taylor.

6. ICA & Expectation Maximization (EM), Presentation6_LML.ppt , 1 July 2008, Eleftherios Garyfallidis.

Up to this point the lectures were based to the presentations of ML course given by Zoubin Ghahramani at the department of engineering, University of Cambridge, http://learning.eng.cam.ac.uk/zoubin/ml06/index.html .

7. Graphical Models 1, 8 July 2008, Ian Nimmo-Smith.

8. Graphical Models 2, 13 January 2009. Ian Nimmo-Smith.

9. Markov Chain Monte Carlo, 20 January 2009, Eleftherios Garyfallidis.

10. ???, 27 January 2009, Hamed Nili.

Proposed next topics :

Non-parametric Methods ( Kernel density estimators, nearest-neighbour methods).
Gaussian Processes.
Sparse Kernel Machines (support vector machines (SVM))
Sparse Kernel Machines 2 (relevance vector machines (RVM))
Boosting.
Overview of clustering methods ( k-means, EM, hierarchical clustering).
Mutual Information with applications to registration and neuronal coding.
Random Field Theory with applications in fMRI.
Artificial Neural Networks from a probabilistic viewpoint.
Machine Learning methods used in SPM.
Machine Learning methods used in FSL.
Signal processing basics.
Fourier Transform.
Wavelets.
Spherical Harmonics.
Spherical Deconvolution.
SNR in MRI experiments.
Variational approximations (KL divergences, mean field, expectation propagation).
Model comparison (Bayes factors, Occam's razor, BIC, Laplace approximations).
Reinforcement Learning, Decision Making and MDPs (value functions, value iteration, policy iteration,
- Bellman equations, Q-learning, Bayesian decision theory

Books

1. Pattern Recognition and Machine Learning, C. M. Bishop, 2006. (Copy in our Library)

2. Information Theory and Learning Algorithms, D. J. C. Mackay, 2003. (Available online)

3. Netlab Algorithms for Pattern Recognition, I. T. Nabney, 2001.

4. Gaussian Processes for Machine Learning, C. E. Rasmussen and C. K. I. Williams, 2006. (Available online)

5. Kernel Methods for Pattern Analysis, Shawe-Taylor and Cristianini, 2004.

Reading

EM

An online demo with mixtures of lines or mixtures of gaussians. http://lcn.epfl.ch/tutorial/english/gaussian/html/

ICA

An online demonstration of the concept in http://www.cis.hut.fi/projects/ica/icademo/

A tutorial is given at http://www.cs.helsinki.fi/u/ahyvarin/papers/NN00new.pdf

A maximum likelihood algorithm for ICA http://www.inference.phy.cam.ac.uk/mackay/ica.pdf

ICA vs PCA

A simple graphical representation of the differences is given in http://genlab.tudelft.nl/~dick/cvonline/ica/node3.html

MCMC

Christophe Andrieu, Nando de Freitas, Arnaud Doucet and Michael I. Jordan. (2003) An Introduction to MCMC for Machine Learning. Machine Learning, 50, 5–43, 2003.

Reversible Jump Markov Chain Monte Carlo (RJMCMC)