Abstract
Neighborhood preserving embedding (NPE) is a classical and very promising supervised dimensional reduction (DR) technique based on a linear graph, which preserves the local neighborhood relations of the data points. However, NPE uses the K nearest neighbor (KNN) criteria for constructing an adjacent graph which makes it more sensitive to neighborhood size. In this article, we propose a novel DR method called weighted neighborhood preserving ensemble embedding (WNPEE). Unlike NPE, the proposed WNPEE constructs an ensemble of adjacent graphs with the number of nearest neighbors varying. With this graph ensemble building, WNPEE can obtain the low-dimensional projections with optimal embedded graph pursuing in a joint optimization manner. WNPEE can be applied in many machine learning fields, such as object recognition, data classification, signal processing, text categorization, and various deep learning tasks. Extensive experiments on Olivetti Research Laboratory (ORL), Georgia Tech, Carnegie Mellon University-Pose and Illumination Images (CMU PIE) and Yale, four face databases demonstrate that WNPEE achieves a competitive and better recognition rate than NPE and other comparative DR methods. Additionally, the proposed WNPEE achieves much lower sensitivity to the neighborhood size parameter as compared to the traditional NPE method while preserving more of the local manifold structure of the high-dimensional data.
Highlights
In many machine learning applications, such as data classification [1,2], face recognition [3,4], signal processing [5,6], and text categorization [7,8], the input data is usually high-dimensional which makes the calculations too complex, as well as requiring more computational time
By computing the value of αk we can iteratively obtain the weights on ensemble of adjacent graphs. Using this graph ensemble building, our proposed weighted neighborhood preserving ensemble embedding (WNPEE) dimensionality reduction (DR) algorithm can obtain the low-dimensional projections with optimal embedded graph pursuing in a joint optimization way
We proposed a new graph-based DR algorithm, called weighted neighborhood preserving ensemble embedding (WNPEE)
Summary
In many machine learning applications, such as data classification [1,2], face recognition [3,4], signal processing [5,6], and text categorization [7,8], the input data is usually high-dimensional which makes the calculations too complex, as well as requiring more computational time. Principal component analysis (PCA) [10] and linear discriminant analysis (LDA) [11] are two well-known linear subspace learning DR approaches. Both PCA and LDA assumes that the sample lies on a linear embedded manifold. PCA preserves the global geometrical information whereas LDA tries to preserve the global discriminant information of the high-dimensional data points. As these both aforementioned linear methods are very simple and convenient to use, they still fail to exploit
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have