A fully robust PARAFAC method for analyzing fluorescence data

Engelen, Sanne; Frosch, Stina; Jørgensen, Bo

doi:10.1002/cem.1208

Cited by 30 publications

(28 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In addition we observe that the AROFAC2 loadings better fulfill the nonnegativity constraints even though they were not enforced explicitly. Furthermore, the peak of the fifth component around 315 nm in both excitation and emission spectra which can be attributed to Rayleigh scatter in all samples [20,11] is sharper and thus in better agreement with its expected shape when identified by AROFAC2.…”

Section: Methodssupporting

confidence: 68%

“…The mathematical theory concerning CP-decompositions of tensors which are not matrices is only partly understood; also, while there exist several methods to calculate the CP-decomposition of a tensor [9,10], they are extrinsical in the sense that a structureagnostic loss function is optimized and also highly sensitive to outliers or non-Gaussian noise -problems which have been heuristically attempted to cope with (e.g. [11]). Moreover, determining the rank of a noisy tensor remains a problematic task despite the existence of heuristics [12].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Approximate rank-detecting factorization of low-rank tensors

Király

Ziehe

2013

2013 IEEE International Conference on Acoustics, Speech and Signal Processing

View full text Add to dashboard Cite

We present an algorithm, AROFAC2, which detects the (CP-)rank of a degree 3 tensor and calculates its factorization into rank-one components. We provide generative conditions for the algorithm to work and demonstrate on both synthetic and real world data that AROFAC2 is a potentially outperforming alternative to the gold standard PARAFAC over which it has the advantages that it can intrinsically detect the true rank, avoids spurious components, and is stable with respect to outliers and non-Gaussian noise.

show abstract

Section: Methodssupporting

confidence: 68%

Section: Introductionmentioning

confidence: 99%

Approximate rank-detecting factorization of low-rank tensors

Király

Ziehe

2013

2013 IEEE International Conference on Acoustics, Speech and Signal Processing

View full text Add to dashboard Cite

show abstract

“…For a review of the common robust methods used in data analysis, see [18,19]. Implementing ROBPCA for scatter identification was described in [9,20]; however, the fact that the method appeared computationally intensive when analyzing vast data structures encouraged a search for the faster alternative.…”

Section: Pca Robpca and S-pcamentioning

confidence: 99%

Automatic scatter detection in fluorescence landscapes by means of spherical principal component analysis

Kotwa¹,

Jørgensen²,

Brockhoff³

et al. 2013

Journal of Chemometrics

Self Cite

View full text Add to dashboard Cite

In this paper, we introduce a new method, based on spherical principal component analysis (S-PCA), for the identification of Rayleigh and Raman scatters in fluorescence excitation-emission data. These scatters should be found and eliminated as a prestep before fitting parallel factor analysis models to the data, in order to avoid model degeneracies. The work is inspired and based on a previous research, where scatter removal was automatic (based on a robust version of PCA called ROBPCA) and required no visual data inspection but appeared to be computationally intensive. To overcome this drawback, we implement the fast S-PCA in the scatter identification routine. Moreover, an additional pattern interpolation step that complements the method, based on robust regression, will be applied. In this way, substantial time savings are gained, and the user's engagement is restricted to a minimum, which might be beneficial for certain applications. We conclude that the subsequent parallel factor analysis models fitted to excitation-emission data after scatter identification based on either ROBPCA or S-PCA are comparable; however, the modified method based on S-PCA clearly outperforms the original approach in relation to computational time. Figure 8. Parallel factor analysis resolved spectra for the North Sea data: (a) spherical principle component analysis (S-PCA) with correction and unimodality constraint, (b) S-PCA with correction and nonnegativity constraint, and (c) S-PCA with correction and unimodality and nonnegativity constraints.

show abstract

“…In this section, we apply our new method for incomplete data to the Dorrit dataset, as considered in . The dataset contains I = 27 mixtures of four known fluorophores: phenylanaline, 3,4‐dihydroxyphenylalanine, 1,4‐dihydroxybenzene and tryptophan.…”

Section: Dorrit Datamentioning

confidence: 99%

“…In this setting, outliers are described as matrices that have a deviating profile compared towards the other ones. This robust PARAFAC method is also a very important component of the approach developed in , which can also handle scatter, which is a type of contamination that affects all samples.…”

Section: Introductionmentioning

confidence: 99%

Robust PARAFAC for incomplete data

Hubert

Kerckhoven

Verdonck

2012

Journal of Chemometrics

View full text Add to dashboard Cite

Different methods exist to explore multiway data. In this article, we focus on the widely used PARAFAC (parallel factor analysis) model, which expresses multiway data in a more compact way without ignoring the underlying complex structure. An alternating least squares procedure is typically used to fit the PARAFAC model. It is, however, well known that least squares techniques are very sensitive to outliers, and hence, the PARAFAC model as a whole is a nonrobust method. Therefore a robust alternative, which can deal with fully observed data possibly contaminated by outlying samples, has already been proposed in literature. In this paper, we present an approach to perform PARAFAC on data that contain both outlying cases and missing elements. A simulation study shows the good performance of our methodology. In particular, we can apply our method on a dataset in which scattering is detected and replaced with missing values. This is illustrated on a real data example. Copyright © 2012 John Wiley & Sons, Ltd.

show abstract

A fully robust PARAFAC method for analyzing fluorescence data

Cited by 30 publications

References 26 publications

Approximate rank-detecting factorization of low-rank tensors

Approximate rank-detecting factorization of low-rank tensors

Automatic scatter detection in fluorescence landscapes by means of spherical principal component analysis

Robust PARAFAC for incomplete data

Contact Info

Product

Resources

About