Discriminative and domain invariant subspace alignment for visual tasks

Rezaei, Samaneh; Tahmoresnezhad, Jafar

doi:10.1007/s42044-019-00037-y

Cited by 14 publications

(10 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…DISA 36 is introduced to find a latent feature subspace for both of the source and the target domains. Give a source domain DS=true{xis, yistrue}i=1ns and a target domain DT=true{xittrue}i=1nt where x i s ∈ X s is a sample of the source domain; y i s is the label of the source sample; n s is the number of the labeled source samples; similarly, x i t ∈ X t is a sample of the target domain; n t is the number of the samples in the target domain.…”

Section: The Methods Of Disamentioning

confidence: 99%

“…where bold-italicSω=∑c=1Cnsctrue(msc−truem¯strue)true(msc−truem¯strue)normalT; bold-italicSb=∑c=1Cbold-italicXscbold-italicHsctrue(bold-italicXsctrue)normalT; bold-italicSt=bold-italicXtbold-italicHtbold-italicXtnormalT; ∑c=0CMc for c = 0 computes MMD coefficient matrix for marginal distribution; for c = 1 through C , computes MMD coefficient matrix in conditional distribution mode; P = [ P 1 P 2 ] T ∈ R 2 m × k , and k is the number of latent subspace dimensions; β , α , γ , λ are the parameters of DISA model. The detailed steps of the DISA method can be found in reference 36 .…”

Section: The Methods Of Disamentioning

confidence: 99%

“…26 drew support from the JDA framework to identify gear and rolling bearing faults in the case of domain shift, which achieved accuracies of more than 96%. In order to find a more satisfactory transfer effect, a new domain adaptation method, which is called discriminative and domain invariant subspace alignment (DISA), was proposed in reference 36 . It is proved to be superior to other domain adaptation methods, such as SA, JDA, VDA, CLGA, etc.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

An intelligent fault diagnosis method based on domain adaptation for rolling bearings under variable load conditions

Zhang

Qin

et al. 2021

Proceedings of the Institution of Mechanical Engineers, Part C:

View full text Add to dashboard Cite

To identify rolling bearing faults under variable load conditions, a method named DISA-KNN is proposed in this paper, which is based on the strategy of feature extraction-domain adaptation-classification. To be specific, the time-domain and frequency-domain indicators are used for feature extraction. Discriminative and domain invariant subspace alignment (DISA) is used to minimize the data distributions’ discrepancies between the training data (source domain) and testing data (target domain). K-nearest neighbor (KNN) is applied to identify rolling bearing faults. DISA-KNN’s validation is proved by the experimental signal collected under different load conditions. The identification accuracies obtained by the DISA-KNN method are more than 90% on four datasets, including one dataset with 99.5% accuracy. The strength of the proposed method is further highlighted by comparisons with the other 8 methods. These results reveal that the proposed method is promising for the rolling bearing fault diagnosis in real rotating machinery.

show abstract

Section: The Methods Of Disamentioning

confidence: 99%

Section: The Methods Of Disamentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

An intelligent fault diagnosis method based on domain adaptation for rolling bearings under variable load conditions

Zhang

Qin

et al. 2021

Proceedings of the Institution of Mechanical Engineers, Part C:

View full text Add to dashboard Cite

show abstract

“…At the local step, CLGA uses both the class discrimination information and data geometric structures. Discriminative and domain invariant subspace alignment for visual tasks (DISA) [33] aims to embed the various domains data into the relevant feature spaces. DISA globally matches both domains through the distribution divergence minimiziation between domains.…”

Section: Related Workmentioning

confidence: 99%

“…We compare DAB with two baseline ML methods, i.e., NN, FLDA and eight novel DA methods according to the mentioned datasets ( joint distribution adaptation (JDA) [41], transfer joint matching (TJM) [42], discriminative transfer subspace learning via low-rank and sparse representation (LRSR) [43], JACRL [26], VDA [31], CLGA [32], DISA [33] and discriminative joint probability MMD for DA (JPDA) [44]). Since these methods are considered as dimensionality reduction approaches, we train a prediction function on the labeled training data (i.e., NN and SVM classifier), and then apply it on test data to estimate the labels of target domain.…”

Section: Comparison Baselinesmentioning

confidence: 99%

Domain Adaptation via Bregman divergence minimization

2021

Self Cite

View full text Add to dashboard Cite

In recent years, the Fisher linear discriminant analysis (FLDA) based classification models are among the most successful approaches and have shown effective performance in different classification tasks. However, when the learning data (source domain) have a different distribution compared with the testing data (target domain), the FLDA-based models may not work well, and the performance degrades, dramatically. To face this issue, we offer an optimal domain adaptation via Bregman divergence minimization (DAB) approach, in which the discriminative features of source and target domains are simultaneously learned via domain invariant representation. DAB is designed based on the constraints of FLDA, with the aim of the coupled marginal and conditional distribution shifts adaptation through Bregman divergence minimization. Thus, the resulting representation can show well functionality like FLDA and simultaneously discriminate across various classes, as well. Moreover, our proposed approach can be easily kernelized to deal with nonlinear tasks. Different experiments on various benchmark datasets demonstrate that our DAB can constructively face with the cross domain divergence and outperforms other novel state-of-the-art domain adaptation approaches in crossdistribution domains.

show abstract