Discriminative Semi-supervised Learning in Manifold Subspace for Face Recognition

Vo, Tue-Minh Dinh; Truong, Hung Phuoc; Le, Thai Hoang

doi:10.1007/978-3-319-29236-6_24

Cited by 1 publication

(1 citation statement)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In a recent study, Thang, et al [25] proved that data clustering in manifold learning is better than in linear subspace and proposed a scheme to overcome the limitation of the traditional Graph K-means algorithm called GKM-LC which always ensures that the number of clusters is stable in each iteration. This paper is the extended version of our published paper [26], we propose a new semi-supervised dimensionality reduction algorithm, called Discriminative Semi-supervised Learning in Manifold subspace (DSLM). Our proposed algorithm aims to find a projection which captures not only the discriminant structure inferred from the labeled data but also the intrinsic geometrical structure inferred from the whole training data.…”

Section: Introductionmentioning

confidence: 99%

Face recognition based on LDA in manifold subspace

Truong¹,

Vo²,

Le³

2016

EAI Endorsed Transactions on Context-aware Systems and Applicat

Self Cite

View full text Add to dashboard Cite

Although LDA has many successes in dimensionality reduction and data separation, it also has disadvantages, especially the small sample size problem in training data because the "within-class scatter" matrix may not be accurately estimated. Moreover, this algorithm can only operate correctly with labeled data in supervised learning. In practice, data collection is very huge and labeling data requires high-cost, thus the combination of a part of labeled data and unlabeled data for this algorithm in Manifold subspace is a novelty research. This paper reports a study that propose a semi-supervised method called DSLM, which aims at overcoming all these limitations. The proposed method ensures that the discriminative information of labeled data and the intrinsic geometric structure of data are mapped to new optimal subspace. Results are obtained from the experiments and compared to several related methods showing the effectiveness of our proposed method.

show abstract

Section: Introductionmentioning

confidence: 99%