Research Article  Open Access
Generalized Discriminant Orthogonal Nonnegative Tensor Factorization for Facial Expression Recognition
Abstract
In order to overcome the limitation of traditional nonnegative factorization algorithms, the paper presents a generalized discriminant orthogonal nonnegative tensor factorization algorithm. At first, the algorithm takes the orthogonal constraint into account to ensure the nonnegativity of the lowdimensional features. Furthermore, the discriminant constraint is imposed on lowdimensional weights to strengthen the discriminant capability of the lowdimensional features. The experiments on facial expression recognition have demonstrated that the algorithm is superior to other nonnegative factorization algorithms.
1. Introduction
Over the past few years, the nonnegative matrix factorization algorithm (NMF) [1] and its variants have proven to be useful for several problems, especially in facial image characterization and representation problems [2–8]. The idea of nonnegative factorization is partly motivated by the biological fact that the firing rates in visual perception neurons are nonnegative.
However, NMF and its variants have some drawbacks. First of all, NMF requires that all object images should be vectorized in order to find the nonnegative decomposition. This vectorization leads to information loss, since the local structure of the image is lost. Moreover, NMF is not unique [9, 10]. In order to remedy these drawbacks, nonnegative tensor factorization (NTF) has been proposed [11–13]. NTF represents a facial expression database as a threeorder tensor. The tensor representation avoids the vectorization operation and preserves the structure of the data. Under some mild conditions, NTF is unique. Existing NMF and NTF algorithms project data into lowdimensional space with the inverse or pseudoinverse of the basis images, so both of them cannot guarantee the nonnegativity of lowdimensional features, which restricts the application of nonnegative factorization in real world. Furthermore, NTF do not take into account class information in data samples. Actually, it is believed that those features with discriminant constraints are of great importance for pattern recognition. Reference [14] develops a discriminant nonnegative tensor factorization algorithm (DNTF), which adds fisher discriminant constraint into the objective function. But like other discriminant nonnegative matrix factorizations [6, 15–18], DNTF employed discriminant analysis on the representation coefficients and not on the actual features used in the recognition procedure. The actual features used for recognition are derived from the projection of data samples to the bases matrix and only implicitly depend on the representation coefficients.
Based on the above analysis, the paper proposes a generalized discriminant orthogonal nonnegative tensor factorization algorithm (GDONTF), which makes full use of the class information and imposes the orthogonal constraint to the objective function. The algorithm not only guarantees the nonnegativity of lowdimensional features, but also generalizes discriminant constraints to lowdimension features. The experiments on facial expression recognition indicate that GDONTF achieves better performance than other nonnegative factorization algorithms.
2. Generalized Discriminant Orthogonal NonNegative Tensor Factorization
Consider an order tensor , every data sample is an order tensor; that is, , in which , is the dimensionality and is the number of data set. The data set is divided into classes. Data samples belonging to class denote ; the number of data samples in is . In order to guarantee the nonnegativity of lowdimensional features and take use of the class information, we propose generalized discriminant orthogonal nonnegative tensor factorization algorithm; the objective function of which is defined as follow:
In which, , , , , is the identity matrix and and are the within and betweenclass scatter matrices of the lowdimensional features, respectively. Because , lowdimensional features can be computed as follows: where the basis matrix . Let be the lowdimensional features of the sample ; then the feature matrix consists of all lowdimensional features, is the low dimensionality of samples, and is the number of all samples. Actually, the separability of the weight coefficient has nothing to do with the recognition accuracy, while the class separability of the lowdimensional features has a great influence on the recognition accuracy. Consequently, the within and betweenclass scatter matrices are defined as follows: where is the mean of the lowdimensional features in the class and is the mean of all lowdimensional features. The objective function in (1) can be written as the following optimization problem:
Since the basis matrix consists of the projection matrices , , we solve the projection matrices , , and the weight matrix , respectively, to deal with the optimization problem (4). First of all, we formulate the Lagrange multipliers out of the constrained optimization problem in (4): where .
Take the derivative of with respect to and , ; we have
Set (6) and (7) to zeros; we get
Left multiply both side of (8) by ; we immediately have Therefore, the update rule for is
The gradient is given by
Because , we can get Let and ; we have Since
We have Similarly, we have To solve the weight matrix , the objective function is The gradient functon is where .
Consequently, the update rules of are
3. Experiments
We have conducted facial expression recognition in order to compare the GDONTF with other algorithms such as NMFOS [19], DNMF [6], FisherNMF [16], and DNTF [14]. Because these algorithms calculate lowdimension features in iteration form, the iteration number is 100. For NMFOS and GDONTF, . in DNMF and in FisherNMF. All lowdimension features are classified by SVM with linear kernel.
The database used for the facial expression recognition experiments is Jaff facial expression database [20]. The database contains 213 images of ten Japanese women. Each person has two to four images for each of the seven expressions: neutral, happy, sad, surprise, anger, disgust, and fear. Each image is resized into . A few examples are shown in Figure 1. We randomly select 20 images from each expression for training; the rest is used for testing. The recognition rates with various dimensionalities of different algorithms are shown in Figure 2. Table 1 shows the best recognition rates of the above algorithms. Because NMF is unsupervised learning algorithm, it has the lowest recognition rates. DNMF and FisherNMF have better recognition rates with supervised learning. It is interesting that NMFOS is superior to DNMF and FisherNMF when the feature dimensionality is from 16 to 160 and is better than DNTF when the feature dimensionality is from 16 to 40, which also illustrates the validity of the orthogonal constraint. It is obvious that GDONTF outperforms other algorithms and the best recognition rate is up to 97.07%.

4. Conclusion
In this paper, a generalized discriminant orthogonal nonnegative tensor factorization algorithm is proposed considering the orthogonal constraint and the discriminant constraint. For the algorithm, the nonnegativity of the lowdimensional features is preserved due to the orthogonal constraint for either training samples or testing samples. In order to enhance the recognition accuracy, the discriminant is conducted on lowdimensional features instead of the weight coefficient of the basis images. The experiments also validate the performance of the algorithm.
Conflict of Interests
The authors declare that there is no conflict of interests regarding the publication of this paper.
References
 D. D. Lee and H. S. Seung, “Learning the parts of objects by nonnegative matrix factorization,” Nature, vol. 401, no. 6755, pp. 788–791, 1999. View at: Publisher Site  Google Scholar
 I. Buciu and I. Pitas, “Application of nonnegative and local non negative matrix factorization to facial expression recognition,” in Proceedings of the 17th International Conference on Pattern Recognition (ICPR '04), pp. 288–291, August 2004. View at: Publisher Site  Google Scholar
 S. Zafeiriou, A. Tefas, I. Buciu, and I. Pitas, “Exploiting discriminant information in nonnegative matrix factorization with application to frontal face verification,” IEEE Transactions on Neural Networks, vol. 17, no. 3, pp. 683–695, 2006. View at: Publisher Site  Google Scholar
 C.J. Lin, “Projected gradient methods for nonnegative matrix factorization,” Neural Computation, vol. 19, no. 10, pp. 2756–2779, 2007. View at: Publisher Site  Google Scholar
 I. Kotsia, S. Zafeiriou, and I. Pitas, “Novel discriminant nonnegative matrix factorization algorithm with applications to facial image characterization problems,” IEEE Transactions on Information Forensics and Security, vol. 2, no. 3, pp. 588–595, 2007. View at: Publisher Site  Google Scholar
 B. Ioan and N. Ioan, “Nonnegative matrix factorization methods for face recognition under extreme lighting variations,” in Proceedings of the International Symposium on Signals, Circuits and Systems (ISSCS '09), Iasi, Romania, July 2009. View at: Publisher Site  Google Scholar
 J. Lu and Y.P. Tan, “Doubly weighted nonnegative matrix factorization for imbalanced face recognition,” in Proceedings of the IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP '09), pp. 877–880, IEEE Computer Society, Taipei, Taiwan, April 2009. View at: Publisher Site  Google Scholar
 H. Tong and C.Y. Lin, “Nonnegative residual matrix factorization: problem definition, fast solutions, and applications,” Statistical Analysis and Data Mining, vol. 5, no. 1, pp. 3–15, 2012. View at: Publisher Site  Google Scholar
 D. Donoho and V. Stodden, “When does nonnegative matrix factorization give a correct decomposition into parts,” Advances in Neural Information Processing Systems, vol. 16, pp. 1141–1148, 2004. View at: Google Scholar
 J. Liu, J. Liu, P. Wonka, and J. Ye, “Sparse nonnegative tensor factorization using columnwise coordinate descent,” Pattern Recognition, vol. 45, no. 1, pp. 649–656, 2012. View at: Publisher Site  Google Scholar
 T. Van De Cruys, “A nonnegative tensor factorization model for selectional preference induction,” Natural Language Engineering, vol. 16, no. 4, pp. 417–437, 2010. View at: Publisher Site  Google Scholar
 S. Zafeiriou, “Discriminant nonnegative tensor factorization algorithms,” IEEE Transactions on Neural Networks, vol. 20, no. 2, pp. 217–235, 2009. View at: Publisher Site  Google Scholar
 C. Liu, K. He, J.L. Zhou, and Y.L. Zhu, “Facial Expression recognition based on Orthogonal Nonnegative CP Factorization,” Advanced Materials Research, vol. 143144, pp. 111–115, 2011. View at: Publisher Site  Google Scholar
 S. Zafeiriou, “Discriminant nonnegative tensor factorization algorithms,” IEEE Transactions on Neural Networks, vol. 20, no. 2, pp. 217–235, 2009. View at: Publisher Site  Google Scholar
 I. Buciu and I. Pitas, “NMF, LNMF, and DNMF modeling of neural receptive fields involved in human facial expression perception,” Journal of Visual Communication and Image Representation, vol. 17, no. 5, pp. 958–969, 2006. View at: Publisher Site  Google Scholar
 Y. Wang, Y. Jia, H. U. Changbo, and M. Turk, “Nonnegative matrix factorization framework for face recognition,” International Journal of Pattern Recognition and Artificial Intelligence, vol. 19, no. 4, pp. 495–511, 2005. View at: Publisher Site  Google Scholar
 X. Li and K. Fukui, “Fisher nonnegative matrix factorization with pairwise weighting,” in Proceedings of the IAPR Conference on Machine Vision Applications, MVA Press, Tokyo, Japan, 2007. View at: Google Scholar
 A. Vilamala, P. J. G. Lisboab, S. OrtegaMartorellc, and A. Vellidoa, “Discriminant convex nonnegative matrix factorization for the classification of human brain tumours,” Pattern Recognition Letters, vol. 34, no. 14, pp. 1734–1747, 2013. View at: Google Scholar
 Z. Li, X. Wu, and H. Peng, “Nonnegative matrix factorization on orthogonal subspace,” Pattern Recognition Letters, vol. 31, no. 9, pp. 905–911, 2010. View at: Publisher Site  Google Scholar
 M. J. Lyons, J. Budynek, and S. Akamatsu, “Automatic classification of single facial images,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 21, no. 12, pp. 1357–1362, 1999. View at: Publisher Site  Google Scholar
Copyright
Copyright © 2014 Zhang XiuJun and Liu Chang. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.