A New Covariance Estimate for Bayesian Classifiers in Biometric Recognition

Thomaz, Carlos Eduardo; Gillies, Donald; Feitosa, Raul Queiroz

doi:10.1109/tcsvt.2003.821984

Cited by 79 publications

(60 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…4 SDNMF efficacy initially increases as we partition each class from 2 up to 5 subclasses, where our algorithm attained its best performance, while further partitioning classes results in reduced recognition accuracy. This is attributed to the fact that since training samples per subclass are limited subclass covariance matrices evaluated on few examples are poorly estimated which affects the correctness of the identified projection directions [34], [48].…”

Section: F Object Recognition On Eth-80 Datasetmentioning

confidence: 99%

Projected Gradients for Subclass Discriminant Nonnegative Subspace Learning

Nikitidis

Tefas

Pitas

2014

IEEE Trans. Cybern.

View full text Add to dashboard Cite

Abstract-Current discriminant nonnegative matrix factorization (NMF) methods either do not guarantee convergence to a stationary limit point or assume a compact data distribution inside classes, thus ignoring intra class variance in extracting discriminant data samples representations. To address both limitations, we regard that data inside each class has a multimodal distribution, forming various subclasses and perform optimization using a projected gradients framework to ensure limit point stationarity. The proposed method combines appropriate clustering-based discriminant criteria in the NMF decomposition cost function, in order to find discriminant projections that enhance class separability in the reduced dimensional projection space, thus improving classification performance. The developed algorithms have been applied to facial expression, face and object recognition, and experimental results verified that they successfully identified discriminant parts, thus enhancing recognition performance.

show abstract

Section: F Object Recognition On Eth-80 Datasetmentioning

confidence: 99%

Projected Gradients for Subclass Discriminant Nonnegative Subspace Learning

Nikitidis

Tefas

Pitas

2014

IEEE Trans. Cybern.

View full text Add to dashboard Cite

show abstract

“…We should also note that for the datasets whose number of training observations N is small compared to their dimensionality F (such as the Sheffield and ETH-80 datasets), the computation of the inverse of the MLE of the sample covariance matrix (16) by the EM-based methods, for instance SMDA and EM-MSDA, will be especially problematic (e.g. see [2], [53]). In these cases, we compute the inverse using the eigenvalue decomposition of the sample covariance matrix, keeping only the eigenvalue components whose eigenvalues are above a specific threshold [2].…”

Section: B Evaluationmentioning

confidence: 99%

“…Between FMSDA and EM-MSDA, we observe that the former tends to perform better when the data dimensionality is larger than the number of the samples, and at the same time many subclasses are necessary in order to capture the subclass structure of the data. In these cases, the training samples per subclass are limited and consequently the subclass covariance matrices are poorly estimated [53]. This adversely affects the performance of EM-based methods.…”

Section: B Evaluationmentioning

confidence: 99%

Mixture Subclass Discriminant Analysis Link to Restricted Gaussian Model and Other Generalizations

Gkalelis

Mezaris

Kompatsiaris

et al. 2013

IEEE Trans. Neural Netw. Learning Syst.

View full text Add to dashboard Cite

Abstract-In this paper, a theoretical link between mixture subclass discriminant analysis (MSDA) and restricted Gaussian model is first presented, and then two further discriminant analysis (DA) methods, fractional step MSDA (FSMSDA) and kernel MSDA (KMSDA) are proposed. Linking MSDA to an appropriate Gaussian model allows the derivation of a new DA method under the Expectation Maximization (EM) framework (EM-MSDA), that derives simultaneously the discriminant subspace as well as the maximum likelihood estimates. The two other proposed methods generalize MSDA in order to solve problems inherited from conventional discriminant analysis. FSMSDA solves the subclass separation problem, that is, the situation when the dimensionality of the discriminant subspace is strictly smaller than the rank of the inter-between-subclass scatter matrix. This is done by an appropriate weighting scheme and the utilization of an iterative algorithm for preserving useful discriminant directions. On the other hand, KMSDA uses the kernel trick to separate data with nonlinearly separable subclass structure. Extensive experimentation shows that the proposed methods outperform conventional MSDA and other LDA variants.

show abstract

“…To avoid both critical issues, we have calculated w lda by using a maximum uncertainty LDA-based approach (MLDA) that considers the issue of stabilizing the S w estimate with a multiple of the identity matrix [26,25,27].…”

Section: Linear Discriminant Analysis (Lda)mentioning

confidence: 99%