Home
Resource
Novel clustering and classification algorithms for big data
PDF
PDF format is widely accepted and good for printing.
Plug-in required
PDF-1(772.40 kb)
Citation & Export
View Usage Statistics
Staff View

Citation & Export
Hide

Simple citation

Ghosh, Debopriya. Novel clustering and classification algorithms for big data. Retrieved from https://doi.org/doi:10.7282/t3-74gr-vt97

Export

Click here for information about Citation Management Tools at Rutgers.

Statistics
Hide

Description

TitleNovel clustering and classification algorithms for big data
NameGhosh, Debopriya (author); Katehakis, Michael (chair); Cabrera, Javier (co-chair); BEN-ISRAEL, ADI (internal member); Metaxas, Dmitri (internal member); Papadimitriou, Spyridon (internal member); Williams, Jerome (internal member); Lubomirski, Mariusz (outside member); Rutgers University; Graduate School - Newark
Date Created2020
Other Date2020-05 (degree)
SubjectManagement
Extent1 online resource (xi, 94 pages)
DescriptionThis dissertation studies two important problems that arise in the analysis of Big Data: high dimensionality and massive size of pertinent samples. In Chapter 1, we developed three novel algorithms for clustering and classification of Big Data. First, a novel two-way clustering approach that combines model-based and weighted K-means clustering methods. The two-way approach results in smaller subgroups of binary features that are of size p or less, so that the possible number of patterns is small enough to be efficiently handled by traditional clustering algorithms. This approach can also handle weighted reduced data to scale on massive sample sizes. In Chapter 2, we derived a weighted probabilistic distance-based clustering technique adjusted for cluster sizes. Chapter 3 introduces an ensemble method called Enriched Random Forest for high dimensional data (where n << p, n is the number of observations and p are the features). This algorithm can address situations where the dimension is very high but only a very small fraction of these features is truly informative.
NotePh.D.
NoteIncludes bibliographical references
Genretheses, ETD doctoral
Persistent URLhttps://doi.org/doi:10.7282/t3-74gr-vt97
LanguageEnglish
CollectionGraduate School - Newark Electronic Theses and Dissertations
Organization NameRutgers, The State University of New Jersey
RightsThe author owns the copyright to this work.
Version 8.5.5
Rutgers University Libraries - Copyright ©2024