Last modified by corander@helsinki_fi on 2024/03/27 10:48
Show last authors
1 = Markovian modelling and Bayesian learning, fall 2015 =
2
3 === Lecturer ===
4
5 [[Jukka Corander>>doc:mathstatHenkilokunta.Corander, Jukka]]
6
7 === Scope ===
8
9 5 sp.
10
11 === Type ===
12
13 Advanced studies
14
15 === Prerequisites ===
16
17 Basic calculus, linear algebra, introductory course on probability and statistical inference are absolutely necessary. First course level knowledge on algebra, probability and inference will be recommendable for many parts of the course. Short course synopsis:
18
19 Part 1. Basic properties of discrete-time Markov chains (DTMCs), irreducibility, ergodicity, stationarity, invariant distributions,higher order Markov chains
20
21 Part 2. Statistical inference for DTMCs, maximum likelihood estimation, Bayesian estimation, inference about the order of a DTMC, full likelihood, model averaging, applications to clustering of DNA sequences
22
23 Part 3. Continuous-time Markov chains (CTMCs), basic properties, waiting time distributions, matrix forward equations, generator, absolute vs relative time, maximum likelihood estimation, unidentifiability of root position, applications to DNA and amino acid sequence evolution
24
25 Part 4. Hidden Markov models (HMMs). Basic properties, inference tasks related to smoothing, filtering and prediction, Derin's algorithm, factorization and recursion for HMM calculations, applications to classification and modeling of DNA sequences
26
27 Part 5. Markov random fields and pseudolikelihood inference, variable-order DTMCs (VOMs/VLMCs), sparse higher-order Markov chains (SMCs), inference for VOM/VLMC/SMC models, applications to sequence prediction and classification
28
29 === Lectures ===
30
31 Weeks 44-49, Tuesday 12-14 and Thursday 12-14 in Exactum room B120. Course end on December 4th.
32
33 Anonymous online chat for the course is available at [[http:~~/~~/presemo.helsinki.fi/mmbl2015>>url:http://presemo.helsinki.fi/mmbl2015||shape="rect"]]
34
35 === Exercises ===
36
37 During weeks 45-49 there will be a weekly exercise session in Exactum room C129 on Thursday 14-16. The teacher responsible for the exercise sessions is Alberto Pessia (he has 'at' helsinki dot fi adress). Solutions to each week's exercises must be presented in the sesssion or at least 1 hour beforehand by email to the responsible teacher.
38
39 Exercises for week 45 are available [[here>>url:http://www.helsinki.fi/bsg/filer/Exercises1.pdf||rel="nofollow" shape="rect" class="external-link"]]
40 Exercises for week 46 are available [[here>>url:http://www.helsinki.fi/bsg/filer/Exercises2.pdf||rel="nofollow" shape="rect" class="external-link"]]
41 Exercises for week 47 are available [[here>>url:http://www.helsinki.fi/bsg/filer/Exercises3.pdf||rel="nofollow" shape="rect" class="external-link"]]
42 Exercises for week 48 are available [[here>>url:http://www.helsinki.fi/bsg/filer/Exercises4.pdf||rel="nofollow" shape="rect" class="external-link"]]
43 Exercises for week 49 are available [[here>>url:http://www.helsinki.fi/bsg/filer/Exercises5.pdf||rel="nofollow" shape="rect" class="external-link"]]
44
45 === Exams ===
46
47 To gain the credits from this course, it is necessary to do at least 50% of the exercises and a home exam. Additional solved exercises will yield bonus points for the grade. The home exam will consist of a number of larger assignments that must be returned by March 1st 2016 to the lecturer. Home exam assignments are [[available here>>url:http://www.helsinki.fi/bsg/filer/HomeExam.pdf||shape="rect"]][[.>>url:http://www.helsinki.fi/bsg/filer/HomeExam.pdf||rel="nofollow" shape="rect" class="external-link"]]
48
49 === Preliminary lecture diary ===
50
51 (% class="external-link" %)
52 Week 44:
53 Tue
54 Log-linear expansions of Markov random field probabilities considered during the lecture are shown in detail [[in this document>>url:http://www.helsinki.fi/bsg/filer/loglinExpansions.pdf||shape="rect"]], three examples on why Markovian type models are useful: [[modeling the Ebola outbreak>>url:http://www.sciencemag.org/content/345/6202/1369.abstract||shape="rect"]], [[HMM and tracking of objects (examples starting at 49:30)>>url:http://www.youtube.com/watch?v=jY2E6ExLxaw||shape="rect"]], [[compression of DNA sequence data>>url:http://www.degruyter.com/view/j/sagmb.2014.13.issue-1/sagmb-2013-0031/sagmb-2013-0031.xml||shape="rect"]] ([[backup of the paper if don't have access rights>>url:http://www.helsinki.fi/bsg/filer/BC_reference.pdf||shape="rect"]]), [[some slides>>url:http://www.helsinki.fi/bsg/filer/PhylogeneticsIntro.pptx||shape="rect"]] about modeling neutral evolution, [[Teaser trailer>>url:http://www.helsinki.fi/bsg/filer/Journey2MarkovianLands.pdf||rel="nofollow" shape="rect" class="external-link"]], [[Eye-opener>>url:http://www.helsinki.fi/bsg/filer/SistersParadoxByBayes.pdf||rel="nofollow" shape="rect" class="external-link"]] on conditional probabilities and Bayes' theorem, basic properties of Markov chains. [[This excerpt>>url:http://www.helsinki.fi/bsg/filer/Koski2.pdf||rel="nofollow" shape="rect" class="external-link"]] from the HMM book by T. Koski is mainly used during the lecture and also [[this short excerpt>>url:http://www.helsinki.fi/bsg/filer/IsaacsonBasics.pdf||rel="nofollow" shape="rect" class="external-link"]] on periodicity from the book of Isaacson & Madsen, Markov chains. For further illustrations and mathematical details on Markov chains, see the link to Sirl and Norris in Bibliography.
55 Thu
56 Basic properties of Markov chains continued. To get going with the basics of simulating Markov chains, you might find [[these Matlab codes>>url:http://www-math.bgsu.edu/%7Ezirbel/ap/||rel="nofollow" shape="rect" class="external-link"]] useful.
57 Week 45:
58 Tue
59 Properties of Markov chains continued. [[Derivation of (4.10)>>url:http://www.helsinki.fi/bsg/filer/DerivationOf410.pdf||rel="nofollow" shape="rect" class="external-link"]] on p 164 in the HMM book. Basics of ML and Bayesian learning, see [[this excerpt>>url:http://www.helsinki.fi/bsg/filer/Koski1.pdf||rel="nofollow" shape="rect" class="external-link"]] from the HMM book by T. Koski.
60 Thu
61 Statistical learning for DTMC's, see [[this excerpt>>url:http://www.helsinki.fi/bsg/filer/Koski3.pdf||rel="nofollow" shape="rect" class="external-link"]] from the HMM book by T. Koski. Also, [[this appendix>>url:http://www.helsinki.fi/bsg/filer/KoskiCh3Appendix.pdf||rel="nofollow" shape="rect" class="external-link"]] from the HMM book is useful for refreshing details on various distributions.
62 Week 46:
63 Tue
64
65 Bayesian estimation of DTMC parameters, Continuous-time Markov chains (see the [[e-book by Koski>>url:http://www.ep.liu.se/ea/lsm/2004/001/||rel="nofollow" shape="rect" class="external-link"]]).
66 Week 47:
67 [[Slides for inference about intractable CTMC models>>url:http://www.helsinki.fi/bsg/filer/Introduction2Inference4IntractableMarkovModels.pdf||shape="rect"]], Continuous-time Markov chains (CTMCs) continued, for an application of birth-death processes to modeling the 2014 Ebola epidemic see [[this paper>>url:http://biorxiv.org/content/biorxiv/early/2014/11/06/011171.full.pdf||shape="rect"]], here is also [[an application of CTMCs to infectious disease epidemiology>>url:http://onlinelibrary.wiley.com/doi/10.1111/biom.12040/abstract||rel="nofollow" shape="rect" class="external-link"]]
68
69 basic properties of hidden Markov models, see: [[Ch. 10>>url:http://www.helsinki.fi/bsg/filer/KoskiBookHMM1.pdf||rel="nofollow" shape="rect" class="external-link"]],[[Ch. 12>>url:http://www.helsinki.fi/bsg/filer/KoskiCh12.pdf||rel="nofollow" shape="rect" class="external-link"]],[[Ch. 13>>url:http://www.helsinki.fi/bsg/filer/KoskiBookHMM2.pdf||rel="nofollow" shape="rect" class="external-link"]],[[Ch. 14>>url:http://www.helsinki.fi/bsg/filer/KoskiCh14.pdf||rel="nofollow" shape="rect" class="external-link"]] from the HMM book. [[An example of using HMM in classification>>url:http://dx.doi.org/10.1016/j.jspi.2012.07.013||rel="nofollow" shape="rect" class="external-link"]]. A biological example of the use of HMM is [[here>>url:http://nar.oxfordjournals.org/content/early/2011/11/07/nar.gkr928.full||rel="nofollow" shape="rect" class="external-link"]]. HMMs are also relevant for a multitude of engineering applications, such as dynamic tracking, an excellent technical review of this field by Arnaud Doucet is [[here>>url:http://www.cs.ubc.ca/%7Earnaud/doucet_johansen_tutorialPF.pdf||rel="nofollow" shape="rect" class="external-link"]], another [[excellent review>>url:http://ais.informatik.uni-freiburg.de/teaching/ss10/robotics/slides/stachniss06phd-basictechniques.pdf||rel="nofollow" shape="rect" class="external-link"]] by Cyrill Stachniss and an excellent short introduction by Bryan Minor is [[here>>url:http://eecs.wsu.edu/%7Ecook/ml/presentations/bryan.pdf||rel="nofollow" shape="rect" class="external-link"]].
70 Weeks 48-49:
71
72 [[Slides for introduction to pseudolikelihood inference for Markov random fields>>url:http://www.helsinki.fi/bsg/filer/Pseudolikelihood2015Corander.pdf||shape="rect"]]
73
74 [[A primer on Occham's razor and Bayesian model comparison for Markov chains>>url:http://www.helsinki.fi/bsg/filer/PrimerOnOccham.pdf||rel="nofollow" shape="rect" class="external-link"]], [[Information-theoretic book by D MacKay where Ch 28 contains a detailed explanation of the Occham's razor principle and Bayesian model comparison>>url:http://www.inference.phy.cam.ac.uk/mackay/itila/||rel="nofollow" shape="rect" class="external-link"]], Bayesian learning of the order of a DTMC, see also the use of information-theoretic criteria for choosing the model dimension as explained in this nice [[review paper>>url:http://www.sal.ufl.edu/eel6935/2008/01311138_ModelOrderSelection_Stoica.pdf||rel="nofollow" shape="rect" class="external-link"]], use of Markov chains to clustering of DNA sequences in metagenomics applications, [[paper1>>url:http://www.helsinki.fi/bsg/filer/BC_reference.pdf||rel="nofollow" shape="rect" class="external-link"]] (published in the journal [[Statistical Applications in Genetics and Molecular Biology>>url:http://www.degruyter.com/view/j/sagmb||rel="nofollow" shape="rect" class="external-link"]]), [[paper2>>url:http://nar.oxfordjournals.org/content/40/12/5240.abstract||rel="nofollow" shape="rect" class="external-link"]]
75 CTMS and HMMs continued, Variable Length Markov chains (see the article by Mächler & Buhlmann mentioned in the bibliography), [[Mixture Transition Markov models>>url:http://www.jstor.org/stable/3182795||shape="rect"]] by Adrian Raftery, [[sparse Markov chains>>url:http://onlinelibrary.wiley.com/doi/10.1111/sjos.12053/pdf||rel="nofollow" shape="rect" class="external-link"]]. [[Finite mixture models and EM-algorithm>>url:http://www.helsinki.fi/bsg/filer/Koski_mixturesEM.pdf||rel="nofollow" shape="rect" class="external-link"]], Markov random fields: basics and learning, see this [[tutorial>>url:https://www.youtube.com/watch?v=vRN_j2j-CC4||shape="rect"]] and this [[recent method based on Bayesian pseudolikelihood>>url:http://arxiv.org/pdf/1401.4988v2.pdf||shape="rect"]].
76
77 === Bibliography ===
78
79 Various references will be used during the course. The lecture diary will also include links to some additional materials. Parts of the following books will be considered:
80
81 Timo Koski. Hidden Markov models for bioinformatics. Kluwer, 2001.
82 Timo Koski & John M. Noble. Bayesian networks: An introduction. Wiley, 2009.
83 Timo Koski. Lectures at RNI on Probabilistic Models and Inference for Phylogenetics. Free e-book available [[here>>url:http://www.ep.liu.se/ea/lsm/2004/001/||rel="nofollow" shape="rect" class="external-link"]].
84
85 In addition, we will consider a number of articles & tutorials (articles not directly linked here are generally available form JSTOR collection or are otherwise online):
86
87 Braun, J.V. & Muller, H-G. Statistical methods for DNA sequence segmentation. Statistical Science, 13, 142-162, 1998.
88 Sirl, D. Markov Chains: An Introduction/Review. [[pdf>>url:http://www.maths.uq.edu.au/MASCOS/Markov05/Sirl.pdf||rel="nofollow" shape="rect" class="external-link"]].
89 Norris, J. Markov chains. CUP, [[see online resource>>url:http://www.statslab.cam.ac.uk/%7Ejames/Markov/||rel="nofollow" shape="rect" class="external-link"]].
90 Gu, L. [[Notes on Dirichlet distribution with relatives>>url:http://www.cs.cmu.edu/%7Eepxing/Class/10701-08s/recitation/dirichlet.pdf||rel="nofollow" shape="rect" class="external-link"]]. This document provides a concise recapitulation of some of the central formulas that are needed in the exercises and assignments when doing Bayesian learning. More comprehensive derivations can be found in several books on Bayesian modeling, e.g. in Koski & Noble (2009), which is listed above.
91 Mächler, M. & Buhlmann, P. Variable length Markov chains: Methodology, computing and software. Journal of Computational and Graphical Statistics 13, 435-455, 2004. Preprint available [[here>>url:ftp://stat.ethz.ch/Research-Reports/104.html||rel="nofollow" shape="rect" class="external-link"]]
92 Kass, R.E. & Raftery, A.E. Bayes factors. Journal of the American Statistical Association, 90, 773-795, 1995.
93
94
95
96
97
98 == [[Registration>>url:https://oodi-www.it.helsinki.fi/hy/opintjakstied.jsp?html=1&Tunniste=57059||shape="rect"]] ==
99
100
101 (% style="color: rgb(96,96,96);" %)Did you forget to register? (%%)[[What to do?>>url:https://wiki.helsinki.fi/display/mathstatOpiskelu/Kysymys4||style="text-decoration: underline;" shape="rect"]]
102
103 == Exercises ==
104
105 === Assignments ===
106
107 * Set 1
108 * Set 2
109
110 === Exercise classes ===
111
112 |=(((
113 Group
114 )))|=(((
115 Day
116 )))|=(((
117 Time
118 )))|=(((
119 Room
120 )))|=(% colspan="1" %)(((
121 Instructor
122 )))
123 |(((
124 1.
125 )))|(((
126
127 )))|(((
128
129 )))|(((
130
131 )))|(% colspan="1" %)(((
132
133 )))
134
135 == Course feedback ==
136
137 Course feedback can be given at any point during the course. Click [[here>>url:https://elomake.helsinki.fi/lomakkeet/11954/lomake.html||style="line-height: 1.4285;" shape="rect"]].