Unsupervised feature selection by regularized self-representation

Zhu, Pengfei; Zuo, Wangmeng; Zhang, Lei; Hu, Qinghua; Shiu, Simon

doi:10.1016/j.patcog.2014.08.006

Cited by 272 publications

(81 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Based on dataset labeling, FS methods can be classified as supervised [1], unsupervised [2], or semisupervised. Supervised FS methods are further categorized into filter, wrapper, and embedded methods.…”

Section: Related Workmentioning

confidence: 99%

An online approach for feature selection for classification in big data

Nazar¹,

Senthilkumar²

2017

Turk J Elec Eng & Comp Sci

View full text Add to dashboard Cite

Feature selection (FS), also known as attribute selection, is a process of selection of a subset of relevant features used in model construction. This process or method improves the classification accuracy by removing irrelevant and noisy features. FS is implemented using either batch learning or online learning. Currently, the FS methods are executed in batch learning. Nevertheless, these techniques take longer execution time and require larger storage space to process the entire dataset. Due to the lack of scalability, the batch learning process cannot be used for large data. In the present study, a scalable efficient Online Feature Selection (OFS) approach using the Sparse Gradient (SGr) technique was proposed to select the features from the dataset online. In this approach, the feature weights are proportionally decremented based on the threshold value, which results in attaining zeros for the insignificant features' weights. In order to demonstrate the efficiency of this approach, an extensive set of experiments was conducted using 13 real-world datasets that range from small to large size. The results of the experiments showed an improved classification accuracy of 15%, which is considered to be significant when compared with the existing methods.

show abstract

Section: Related Workmentioning

confidence: 99%

An online approach for feature selection for classification in big data

Nazar¹,

Senthilkumar²

2017

Turk J Elec Eng & Comp Sci

View full text Add to dashboard Cite

show abstract

“…Ant colony optimization (ACO [36]) with some modifications termed ABACO is used for feature selection [10]. Graph regularized Non Matrix Factorization (GNMF) is developed for feature selection [7]. An objective function is defined which finds a subspace such that all samples of the subspace are very far from all other samples.…”

Section: Literature Surveymentioning

confidence: 99%

“…Eliminating redundant, noisy and irrelevant features from datasets is defined as Dimensionality Reduction. Dimensionality reduction is used in face image dataset [6,7,8,9], micro array dataset and speech signals [6,7], digit images [6,7,8], letter images [8,10] for classification or clustering. Feature selection refers to selecting a subset of features from a complete set of features in a dataset.…”

Section: Introductionmentioning

confidence: 99%

“…Feature selection refers to selecting a subset of features from a complete set of features in a dataset. Feature selection is used for the classification of various data sets like Abalone, Glass, Iris, Letter, Shuttle, Spam base, Tae, Vehicle, Waveform, Wine, Wisconsin and Yeast data [7]. Feature selection and dimensionality reduction possess mostly a common goal which is to reduce the number of harmful, irrelevant and noisy features in a dataset for smooth and fast data processing purposes.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A Survey on Feature Selection Algorithms

Kumar¹

2015

IJRITCC

View full text Add to dashboard Cite

Abstract-One major component of machine learning is feature analysis which comprises of mainly two processes: feature selection and fe ature extraction. Due to its applications in several areas including data mining, soft computing and big data analysis, feature selection has got a reasonable importance. This paper presents an introductory concept of feature selection with various inherent approaches. The paper surveys historic developments reported in feature selection with supervised and unsupervised methods. The recent developments with the state of the art in the on-going feature selection algorithms have also been summarized in the paper including their hybridizations.

show abstract

“…Supervised and semi-supervised methods are usually applied on labeled data, while un-supervised method is more appropriate for unlabeled data [3]. However, many real-world applications do not contain any label, hence, the unsupervised feature selection process becoming difficult and hard to achieve [4].…”

Section: Introductionmentioning

confidence: 99%

Unsupervised Feature Selection Method for Intrusion Detection System

Ambusaidi

Nanda

2015

2015 IEEE Trustcom/BigDataSE/Ispa

View full text Add to dashboard Cite

Abstract-This paper considers the feature selection problem for data classification in the absence of data labels. Due to the lack of categorized information in many practical applications, unsupervised feature selection has been proven to be more practically important but at the same time more difficult. It is not an easy task to assess the relevance of a feature or a subset of features when there are no labels available with the data. In this paper, we first propose an unsupervised feature selection algorithm, which is an enhancement over Laplacian score method. We name our algorithm a Modified Laplacian score, M L in short. Specifically, two main phases are involved in M L to complete the selection procedures. In the first phase, the Laplacian score algorithm is applied to select the features that have the best locality preserving power. In the second phase, M L introduces a new Redundancy Penalization (RP) technique based on Mutual Information (MI) to eliminate the redundancy among the selected features. We evaluate our work through applying the proposed unsupervised feature selection algorithm to build an Intrusion Detection System. The effectiveness and the feasibility of the proposed detection system are evaluated using three wellknown intrusion detection datasets: KDD Cup 99, NSL-KDD and Kyoto 2006+ dataset. The evaluation results confirm that our feature selection approach performs better than the Laplacian score method in terms of classification accuracy.

show abstract

Unsupervised feature selection by regularized self-representation

Cited by 272 publications

References 14 publications

An online approach for feature selection for classification in big data

An online approach for feature selection for classification in big data

A Survey on Feature Selection Algorithms

Unsupervised Feature Selection Method for Intrusion Detection System

Contact Info

Product

Resources

About