Mass spectra separation for explosives detection by using probabilistic latent component analysis

Kawaguchi, Yohei; Togami, Masahito; Nagano, H.; Hashimoto, Yuichiro; Sugiyama, Megumi; Takada, Yuki

doi:10.1109/icassp.2012.6288216

Cited by 4 publications

(7 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As our conventional work [ 7], in mass spectra separation domain, the correct solution is likely to a sparse solution in terms of both time direction and mlz direc tion. However.…”

Section: Proposed Me Thodmentioning

confidence: 99%

“…In this section, we explain about the conventional mass spectra sep araion method based on PLCA [ 7]. The PLCA model considers that…”

Section: Problem Stat Ementmentioning

confidence: 99%

“…Thus, by initializing the unknown parameters by ICA and reducing the num ber of iterations of PLCA. we aim to shorten the total calculation 2796 Algorithm 1 PLCA [ 7] 1. Initialization process Set all the unknown parameters to random values.…”

Section: Proposed Me Thodmentioning

confidence: 99%

“…Furthermore. the conventional method based on PLeA [ 7] can solve the uncertainty problem of the number of compounds by using statistical knowledge as sparsity priors. However, the convergence of the method is slow, and the total calculation time is long.…”

Section: Introductionmentioning

confidence: 99%

“…Thus these methods suffer from performance degradation. Recently, there have been several researches that apply non-negative matrix factorization (NMF) [ 5,6] and probabilistic latent component analysis (PLeA) [ 7] to the area of mass spectrometry. These approaches have the desirable feature that the estimated components are guaranteed to be non-negative, and the approaches have the advantage that distortion is not caused by negative values.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

ICA-based acceleration of probabilistic latent component analysis for mass spectrometry-based explosives detection

Kawaguchi

Togami

Nagano

et al. 2013

2013 IEEE International Conference on Acoustics, Speech and Signal Processing

Self Cite

View full text Add to dashboard Cite

We propose a new method to separate mass spectra into compo nents of each chemical compound for explosives detection. The conventional method based on probabilistic latent component anal ysis (PLeA) is effective because the method can solve the prob lems of non-negativity and non-orthogonality by using sparsity of the domain of explosives detection. However, the convergence of the method is slow. and the calculation time is long. In order to solve this problem. the proposed method makes use of independent component analysis (leA) in the initialization process. Experimen tal results indicate that the convergence of the proposed method is accelerated, and total calculation time is decreased.Index Terms-Mass spectrometry. Blind source separation. Probabilistic latent component analysis (PLeA), Independent com ponent analysis (leA), Sparsity INTRODUC TIONThe threat of improvised explosive devices has become a serious problem for all countries because the procedures and recipes for making them are freely available on the Internet. To prevent ter rorist attacks. we have developed a walkthrough portal explosives detector that consists of a high-throughput vapor sampling portal. a high-sensitivity atmospheric pressure chemical ionization source, and a high-selectivity linear ion trap mass spectrometer [ 1]. The mass spectrometer measures the intensity corresponded to the num ber of ions for each mass-to-charge ratio (mlz). The mlz series of intensities are called the mass spectrum. The detector observes the time series of the mass spectra continuously. and it detects charac teristic patterns of explosives traces from the mass spectra data.In mass spectra of the explosives detection system, explosives compounds, other chemical compounds, and the chemical back ground are mixed with each other. Thus. it is necessary to separate the mass spectra into the different compounds. The system does not know what kind of chemical compounds can be measured in advance, and so the task of the system is a blind source separa tion (BSS) problem. There are many researches that employ BSS for mass spectra separation, such as principal component analy sis (peA) [ 2] and independent component analysis (leA) [ 3,4]. Because peA and leA impose the orthogonality and the indepen dence respectively without constraints of non-negativity, and so these methods are not fit to mass spectrometry domain. Thus these methods suffer from performance degradation. Recently, there have been several researches that apply non-negative matrix factorization (NMF) [ 5, 6] and probabilistic latent component analysis (PLeA) 978-1-4799-0356-6/13/$31.00 ©2013 IEEE 2795[ 7] to the area of mass spectrometry. These approaches have the desirable feature that the estimated components are guaranteed to be non-negative, and the approaches have the advantage that distortion is not caused by negative values. Furthermore. the conventional method based on PLeA [ 7] can solve the uncertainty problem of the number of compounds by using statistical knowledge as sparsity priors. ...

show abstract

“…As our conventional work [ 7], in mass spectra separation domain, the correct solution is likely to a sparse solution in terms of both time direction and mlz direc tion. However.…”

Section: Proposed Me Thodmentioning

confidence: 99%

“…In this section, we explain about the conventional mass spectra sep araion method based on PLCA [ 7]. The PLCA model considers that…”

Section: Problem Stat Ementmentioning

confidence: 99%

Section: Proposed Me Thodmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations