Eylem Seç
Combinatorial Inference for Large-Scale Data Analysis
Başlık:
Combinatorial Inference for Large-Scale Data Analysis
Yazar:
Lu, Junwei, author.
ISBN:
9780438047709
Yazar Ek Girişi:
Fiziksel Tanımlama:
1 electronic resource (237 pages)
Genel Not:
Source: Dissertation Abstracts International, Volume: 79-10(E), Section: B.
Advisors: Han Liu; Jianqing Fan Committee members: Jianqing Fan; Han Liu; Mengdi Wang.
Özet:
Problems of inferring the combinatorial structures of networks arise in many real applications ranging from genomic regulatory networks, brain networks to social networks. This poses new and challenging problems on the uncertainty assessment and replicability analysis for statistical methods inferring these network topological structures. This thesis develops combinatorial inference, a new inferential framework for networks, to conduct hypothesis tests and variable selection for combinatorial structures in large-scale graphs.
In the first part of the thesis, we propose a unified inferential method to test hypotheses on the global combinatorial properties of graphical models. We showed that my method works for general monotone graph properties that can be preserved under edge deletion including bipartite, planar, k-colorable, etc. We develop a new combinatorial minimax theory to justify the optimality of the skip-down algorithm. We introduce a new notion of graph packing entropy to sharply characterize the complexity of combinatorial inference problems for graphical models. It can be viewed as the combinatorial counterpart of the famous metric entropy theory on parametric minimax lower bounds.
In the second part of the thesis, we generalize the combinatorial inference for larger family of graphical models. We propose a novel class of dynamic nonparanormal graphical models, which allows us to model high dimensional heavy-tailed systems and the evolution of their latent network structures. Under this model we develop statistical tests for presence of edges both locally at a fixed index value and globally over a range of values. The tests are developed for a high-dimensional regime, are robust to model selection mistakes and do not require commonly assumed minimum signal strength. The testing procedures are based on a high dimensional, debiasing-free moment estimator, which uses a novel kernel smoothed Kendall's tau correlation matrix as an input statistic.
Notlar:
School code: 0181
Tüzel Kişi Ek Girişi:
Mevcut:*
Yer Numarası | Demirbaş Numarası | Shelf Location | Lokasyon / Statüsü / İade Tarihi |
---|---|---|---|
XX(681027.1) | 681027-1001 | Proquest E-Tez Koleksiyonu | Arıyor... |
On Order
Liste seç
Bunu varsayılan liste yap.
Öğeler başarıyla eklendi
Öğeler eklenirken hata oldu. Lütfen tekrar deneyiniz.
:
Select An Item
Data usage warning: You will receive one text message for each title you selected.
Standard text messaging rates apply.