Skip to main content
Login | Suomeksi | På svenska | In English

Quasi-pseudolikelihood in Markov network structure learning

Show simple item record 2016-10-11T08:13:20Z und 2017-10-24T12:22:05Z 2016-10-11T08:13:20Z und 2017-10-24T12:22:05Z 2016-10-11T08:13:20Z
dc.identifier.uri und
dc.title Quasi-pseudolikelihood in Markov network structure learning en
ethesis.discipline Statistics en
ethesis.discipline Tilastotiede fi
ethesis.discipline Statistik sv
ethesis.department Institutionen för matematik och statistik sv
ethesis.department Department of Mathematics and Statistics en
ethesis.department Matematiikan ja tilastotieteen laitos fi
ethesis.faculty Matematisk-naturvetenskapliga fakulteten sv
ethesis.faculty Matemaattis-luonnontieteellinen tiedekunta fi
ethesis.faculty Faculty of Science en
ethesis.faculty.URI Helsingfors universitet sv University of Helsinki en Helsingin yliopisto fi
dct.creator Heikkilä, Mikko
dct.issued 2016
dct.language.ISO639-2 eng
dct.abstract Probabilistic graphical models are a versatile tool for doing statistical inference with complex models. The main impediment for their use, especially with more elaborate models, is the heavy computational cost incurred. The development of approximations that enable the use of graphical models in various tasks while requiring less computational resources is therefore an important area of research. In this thesis, we test one such recently proposed family of approximations, called quasi-pseudolikelihood (QPL). Graphical models come in two main variants: directed models and undirected models, of which the latter are also called Markov networks or Markov random fields. Here we focus solely on the undirected case with continuous valued variables. The specific inference task the QPL approximations target is model structure learning, i.e. learning the model dependence structure from data. In the theoretical part of the thesis, we define the basic concepts that underpin the use of graphical models and derive the general QPL approximation. As a novel contribution, we show that one member of the QPL approximation family is not consistent in the general case: asymptotically, for this QPL version, there exists a case where the learned dependence structure does not converge to the true model structure. In the empirical part of the thesis, we test two members of the QPL family on simulated datasets. We generate datasets from Ising models and Sherrington-Kirkpatrick models and try to learn them using QPL approximations. As a reference method, we use the well-established Graphical lasso (Glasso). Based on our results, the tested QPL approximations work well with relatively sparse dependence structures, while the more densely connected models, especially with weaker interaction strengths, present challenges that call for further research. en
dct.language en
ethesis.language English en
ethesis.language englanti fi
ethesis.language engelska sv
ethesis.thesistype pro gradu-avhandlingar sv
ethesis.thesistype pro gradu -tutkielmat fi
ethesis.thesistype master's thesis en
dct.identifier.urn URN:NBN:fi-fe2017112251832
dc.type.dcmitype Text

Files in this item

Files Size Format View
Pro-gradu_Mikko_Heikkila.pdf 4.532Mb PDF

This item appears in the following Collection(s)

Show simple item record