New Methods of Data Clustering and Classification Based on NMF

Tang, Jie; Ceng, Xinyu; Peng, Bo

doi:10.1109/bcgin.2011.114

Cited by 5 publications

(5 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To examine the results linearly, separate data was manifested. NMF using Multiplicative approach [9] on parts-based data and NMF to new matrix decomposition method reduce the actual data matrix and yield better results [10].…”

Section: Literature Surveymentioning

confidence: 99%

Clustering and Indexing of Multiple Documents Using Feature Extraction Through Apache Hadoop on Big Data

Lydia¹,

Gummadi²,

Varadarajan

et al. 2020

MJCS

View full text Add to dashboard Cite

Bigdata is a challenging field in data processing since the information is retrieved from various search engines through internet. A number of large organizations, that use document clustering,fails in arranging the documents sequentially in their machines. Across the globe, advanced technologyhas contributed to the high speed internet access. But the consequences of useful yet unorganized information in machine files seemto be confused in the retrieval process. Manual ordering of files has its own complications. In this paper, application software like Apache Lucene and Hadoop have taken a lead towards text mining for indexing and parallel implementation of document clustering. In organizations, it identifies the structure of the text data in computer files and its arrangement from files to folders, folders to subfolders, and to higher folders. A deeper analysis of document clustering was performed by considering various efficient algorithms like LSI, SVD and was compared with the newly proposed updated model of Non-Negative Matrix Factorization. The parallel implementation of hadoopdevelopedautomatic clusters for similar documents. MapReduce framework enforced its approach using K-means algorithm for all the incoming documents. The final clusters were automatically organized in folders using Apache Lucene in machines. This model was tested by considering the dataset of Newsgroup20 text documents. Thus this paper determines the implementation of large scale documents using parallel performance of MapReduce and Lucenethat generate automatic arrangement of documents, which reduces the computational time and improves the quick retrieval of documents in any scenario.

show abstract

Section: Literature Surveymentioning

confidence: 99%

Clustering and Indexing of Multiple Documents Using Feature Extraction Through Apache Hadoop on Big Data

Lydia¹,

Gummadi²,

Varadarajan

et al. 2020

MJCS

View full text Add to dashboard Cite

show abstract

“…The columns of K can be regarded as characteristic components of the given data set {Y •,j } j . If these data vectors are input for a classification task, one can use the p correlation values with the column vectors of K as features for constructing the classification scheme, which yields efficient and qualitatively excellent classifications, see [26,30,34].…”

Section: Non-negative Matrix Factorizationmentioning

confidence: 99%

“…Therefore, we have the following Theorem 5 (Surrogate Functional for TV Penalty Term) We consider the cost functional F (K ) := TV(K ) with the total variation defined in (34). Then…”

Section: Definition 7 (Total Variation (Continuous))mentioning

confidence: 99%

See 1 more Smart Citation

A Survey on Surrogate Approaches to Non-negative Matrix Factorization

Fernsel

Maaß

2018

Vietnam J. Math.

View full text Add to dashboard Cite

Motivated by applications in hyperspectral imaging we investigate methods for approximating a high-dimensional non-negative matrix Y by a product of two lower-dimensional, non-negative matrices K and X . This so-called nonnegative matrix factorization is based on defining suitable Tikhonov functionals, which combine a discrepancy measure for Y ≈ KX with penalty terms for enforcing additional properties of K and X . The minimization is based on alternating minimization with respect to K or X , where in each iteration step one replaces the original Tikhonov functional by a locally defined surrogate functional. The choice of surrogate functionals is crucial: It should allow a comparatively simple minimization and simultaneously its first order optimality condition should lead to multiplicative update rules, which automatically preserve non-negativity of the iterates. We review the most standard construction principles for surrogate functionals for Frobenius-norm and Kullback-Leibler discrepancy measures. We extend the known surrogate constructions by a general framework, which allows to add a large variety of penalty terms. The paper finishes by deriving the corresponding alternating minimization schemes explicitely and by applying these methods to MALDI imaging data.

show abstract

“…Xu et al [8] demonstrate that NMF-based indexing outperforms traditional vector space approaches to information retrieval for document clustering on a few benchmark test collections. Tang et al [9] propose new methods of data clustering and classification based on NMF separately. They reduce the dimension of the original term by document matrix and run clustering and classification algorithms on the encoded matrix after NMF processing instead of the original matrix.…”

Section: Introductionmentioning

confidence: 99%

NMF based dimension reduction methods for Turkish text clustering

et al. 2013

View full text Add to dashboard Cite

Conference: IEEE International Symposium on INnovations in Intelligent SysTems and Applications (INISTA) -- Location: BULGARIA -- Date: JUN 19-21, 2013In this work, we analyze the effects of NMF based dimension reduction methods on clustering of Turkish documents by using k-means clustering algorithm. All experiments are conducted on two different datasets that we call Milliyet4c1k and 1150haber. The NMF based dimension reduction methods have two purposes: to reduce the original vector space by transformation and to reduce size and dimension by summarizing original documents. Experimental results show that NMF transformation yields to better clustering results on both datasets. Using k-means on summarized documents produces almost identical result with k-means on original documents. Although using summaries instead of full documents doesn't improve quality of clustering, we show that it significantly reduces the size of the processed data and execution time of k-means clustering algorithm.IEEE; Bulgarian Sci Acad; Bulgarian Acad Sci, Inst Informat & Commun Technologies; IEEE Bulgarian Sectio

show abstract

New Methods of Data Clustering and Classification Based on NMF

Cited by 5 publications

References 4 publications

Clustering and Indexing of Multiple Documents Using Feature Extraction Through Apache Hadoop on Big Data

Clustering and Indexing of Multiple Documents Using Feature Extraction Through Apache Hadoop on Big Data

A Survey on Surrogate Approaches to Non-negative Matrix Factorization

NMF based dimension reduction methods for Turkish text clustering

Contact Info

Product

Resources

About