Stable and orthogonal local discriminant embedding using trace ratio criterion for dimensionality reduction
Stable orthogonal local discriminant embedding (SOLDE) is a recently proposed dimensionality reduction method, in which the similarity, diversity and interclass separability of the data samples are well utilized to obtain a set of orthogonal projection vectors. By combining multiple features of data, it outperforms many prevalent dimensionality reduction methods. However, the orthogonal projection vectors are obtained by a step-by-step procedure, which makes it computationally expensive. By generalizing the objective function of the SOLDE to a trace ratio problem, we propose a stable and orthogonal local discriminant embedding using trace ratio criterion (SOLDE-TR) for dimensionality reduction. An iterative procedure is provided to solve the trace ratio problem, due to which the SOLDE-TR method is always faster than the SOLDE. The projection vectors of the SOLDE-TR will always converge to a global solution, and the performances are always better than that of the SOLDE. Experimental results on two public image databases demonstrate the effectiveness and advantages of the proposed method.
KeywordsTrace ratio criterion Manifold learning Dimensionality reduction Diversity
This paper is supported by National Natural Science Foundation of China (No.61401471 and No.61501471); General Financial from the China Postdoctoral Science Foundation (No.2014M552589) and Special Financial from the China Postdoctoral Science Foundation (No.2015T81114).
- 2.Benmokhtar R, Delhumeau J, Gosselin P H (2013) Efficient supervised dimensionality reduction for image categorization. In: Proceedings of the 2013 IEEE international conference on acoustics, speech and signal processing (ICASSP), pp 2425–2428Google Scholar
- 3.Boughnim N, Marot J, Fossati C, Bourennane S, Guerault F (2013) Fast and improved hand classification using dimensionality reduction and test set reduction. In: Proceedings of the 2013 IEEE international conference on acoustics, speech and signal processing (ICASSP), pp 1971–1975Google Scholar
- 5.Cai D, He X, Zhou K, Han J, Bao H (2007) Locality sensitive discriminant analysis. In: Proceedings of the 20th international joint conference on artificial intelligence, pp 708–713Google Scholar
- 6.Chang X, Ma Z, Yang Y, Zeng Z, Hauptmann AG (2016) Bi-level semantic representation analysis for multimedia event detection. IEEE Trans Cybern 1–18. doi: 10.1109/TCYB.2016.2539546
- 8.Chang X, Yang Y (2016) Semisupervised feature analysis by mining correlations among multiple tasks. IEEE Trans Neural Netw Learn Syst 1–12. doi: 10.1109/TNNLS.2016.2582746
- 9.Chang X, Yang Y, Long G, Zhang C, Hauptmann A G (2016) Dynamic concept composition for zero-example event detection. In: AAAIGoogle Scholar
- 10.Chang X, Yu YL, Yang Y, Xing EP (2016) Semantic pooling for complex event analysis in untrimmed videos. IEEE Trans Pattern Anal Mach Intell (99). doi: 10.1109/TCYB.2015.2479645
- 18.Luo M, Chang X, Nie L, Yang Y, Hauptmann AG, Zheng Q (2017) An adaptive semisupervised feature analysis for video semantic recognition. IEEE Trans Cybern. doi: 10.1109/TCYB.2017.2647904
- 19.Parisotto E, Ghassabeh Y A, Freydoonnejad S, Rudzicz F (2015) Eeg dimensionality reduction in automatic identification of synonymy. In: 2015 IEEE international conference on acoustics, speech and signal processing (ICASSP), pp 847–851Google Scholar
- 21.Wang H, Yan S, Xu D, Tang X, Huang T (2007) Trace ratio vs. ratio trace for dimensionality reduction. In: Proceedings of the 2007 IEEE conference on computer vision and pattern recognition, pp 1–8Google Scholar
- 26.Zoidi O, Nikolaidis N, Pitas I (2014) Semi-supervised dimensionality reduction on data with multiple representations for label propagation on facial images. In: Proceedings of the 2014 IEEE international conference on acoustics, speech and signal processing (ICASSP), pp 6019–6023Google Scholar