Elsevier

Pattern Recognition

Volume 37, Issue 5, May 2004, Pages 1081-1083
Pattern Recognition

Rapid and Brief Communication
Subpattern-based principle component analysis

https://doi.org/10.1016/j.patcog.2003.09.004Get rights and content

Abstract

We propose a subpattern-based principle component analysis (SpPCA). The traditional PCA operates directly on a whole pattern represented as a vector and acquires a set of projection vectors to extract global features from given training patterns. SpPCA operates instead directly on a set of partitioned subpatterns of the original pattern and acquires a set of projection sub-vectors for each partition to extract corresponding local sub-features and then synthesizes them into global features for subsequent classification. The experimental results show that the proposed SpPCA has (much) better classification performances on all the real-life benchmark datasets than PCA.

Introduction

The traditional PCA [1] is a very effective approach of extracting features and has successfully been applied in pattern recognition such as face classification [2]. It operates directly on whole patterns represented as (feature) vectors to extract so-needed global features for subsequent classification by a set of previously found global projectors from a given training pattern set, whose aim is to maximally preserve original pattern information after extracting features, i.e., reducing dimensionality. In this paper, we develop another PCA operating directly on subpatterns rather than on whole pattern. These subpatterns are formed via a partition for an original whole pattern and utilized to compose multiple training subpattern sets for the original training pattern set. In this way, SpPCA can independently be performed on individual training subpattern sets and finds corresponding local projection sub-vectors, and then uses them to extract local sub-features from any given pattern. Afterwards, these extracted sub-features from individual subpatterns are synthesized into a global feature of the original whole pattern for subsequent classification.

Section snippets

Proposed SpPCA

SpPCA includes two steps. In the first step, an original whole pattern denoted by a vector is partitioned into a set of equally-sized subpatterns in non-overlapping ways and then all those subpatterns sharing the same original feature components are respectively collected from the training set to compose corresponding training subpattern sets. Secondly, PCA is performed on each of such subpattern sets. More specifically, we are given a set of training patterns X={X1,X2,…,XN} with each column

Experimental results

The experimental datasets are from publicly attainable 8 benchmark datasets including a derived one. For producing as many partitions as possible, all dimensions of patterns are not less than 12. Here we just give brief experimental conditions because of space limited.

(1) ORL face data1 (40 classes, 10 image patterns each class, 5 patterns each class for training and the rest for testing); (2) Letter data2

Conclusions

We proposed a powerful new approach of extracting features based on PCA and pattern partitioning technique, and made PCA become a special case of the proposed SpPCA. By extracting local sub-features from partitioned subpattern sets and then synthesizing them into global features for subsequent classification. As a result, classification accuracies incorporating the NN rule on all datasets employed here are greatly improved.

Acknowledgements

We thank National Science Foundations of China and of Jiangsu under Grant Nos. 60271017 and BK2002092, the “QingLan” Project Foundation of Jiangsu Province and the Returnee Foundation of China Scholarship Council for partial supports respectively.

References (3)

  • T. Young et al.

    Handbook of Pattern Recognition and Image Processing

    (1986)
There are more references available in the full text version of this article.

Cited by (169)

  • Three-stage circular efficiency evaluation of agricultural food production, food consumption, and food waste recycling in EU countries

    2022, Journal of Cleaner Production
    Citation Excerpt :

    Tone (2001), Slack-Based Measure (SBM) model applied the slack of variables as the basic measurement between the input and output variables, adopting the estimation of the non-radial method with a scalar representing the efficiency of SBM. In addition to the above-mentioned CCR, BCC, and SBM DEA models, there are different models as Chen and Zhu (2004), Kao and Hwang (2008), and Kao (2009) studied the two-stage DEA model. Färe et al. (2007) proposed Network Data Envelopment Analysis, (Network DEA), claiming that the production process is composed of many production technologies and regarded it as Sub-DMU.

  • Dimensionality reduction using PCAs in feature partitioning framework

    2022, Statistical Modeling in Machine Learning: Concepts and Applications
View all citing articles on Scopus
View full text