Beta 1


Title Inference in complex networks
Author Herlau, Tue
Supervisor Hansen, Lars Kai (Cognitive Systems, Department of Informatics and Mathematical Modeling, Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark)
Mørup, Morten (Cognitive Systems, Department of Informatics and Mathematical Modeling, Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark)
Schmidt, Mikkel Nørgaard (Cognitive Systems, Department of Informatics and Mathematical Modeling, Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark)
Institution Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark
Thesis level Master's thesis
Year 2011
Abstract In this thesis, we examine methods for inferring structure in network data within the nonparametric bayesian paradigm. We implement a common method for clustering, the Infinite Relational Model, and develop a novel non-parametric model which can infer if hiarchical structure is present in the data, and if not, reduce to the IRM model. These methods are applied to fMRI data. In answering the problem of when structure can be inferred, we examine the replicate symmetry method from statistical physics, and propose a reformulation as plausible inferense over messages in belief propagation.
Imprint Technical University of Denmark (DTU) : Kgs. Lyngby, Denmark
Series IMM-M.Sc.-2011-27
Fulltext
Original PDF ep11_27.pdf (2.68 MB)
Admin Creation date: 2011-05-05    Update date: 2011-05-05    Source: dtu    ID: 276568    Original MXD