Minimax sparse detection based on one-class classifiers

Suleiman, Raja Fazliza Raja; Mary, David; Ferrari, A.

doi:10.1109/icassp.2013.6638726

Cited by 7 publications

(17 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…detection power) to implement a test of the form (19) with R = L (i.e the concatenation of all possible alternatives). This is because the PFA may increase wildly with the number of alternatives [26]. Hence, dictionaries with reduced dimensions such as those cited above are very useful.…”

Section: Design Strategies For Rmentioning

confidence: 98%

“…In such cases, R is often simply taken as the mean or the first singular vector obtained by a SVD of the data set [30], or it can be optimized using now classical sparse dictionary learning techniques, for instance K-SVD [25]. Another set of approaches called minimax seeks a dictionary of reduced dimension, which maximizes the worst probability of detection under H1 [26,31].…”

Section: Design Strategies For Rmentioning

confidence: 99%

“…Next section compares the 1D (13) and 3D (19) one-sparse GLR tests for the K-SVD [25] and the minimax approaches of [26].…”

Section: Design Strategies For Rmentioning

confidence: 99%

“…We consider that the spectral pattern to be detected is not known, but that we dispose of a set of training signals that represents well the variety of features to be detected. The spectral dictionary used in the proposed test is made of atoms that are learned using dictionary learning techniques [25,26], and further spread by the SSF and WSF, leading to 3D atoms. Note that we do not attempt to first deconvolve the data, and then detect.…”

Section: Introduction and Previous Workmentioning

confidence: 99%

See 3 more Smart Citations

Constrained likelihood ratios for detecting sparse signals in highly noisy 3D data

Paris

Suleiman

Mary

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 method aimed at detecting weak, sparse signals in highly noisy three-dimensional (3D) data. 3D data sets usually combine two spatial directions x and y (e.g. image or video frame dimensions) with an additional direction λ (e.g. temporal, spectral or energy dimension). Such data most often suffer from information leakage caused by the acquisition system's point spread functions, which may be different and variable in the three dimensions. The proposed test is based on dedicated 3D dictionaries, and exploits both the sparsity of the data along the λ direction and the information spread in the three dimensions. Numerical results are shown in the context of astrophysical hyperspectral data, for which the proposed 3D model substantially improves over 1D detection approaches.

show abstract

Section: Design Strategies For Rmentioning

confidence: 98%

Section: Design Strategies For Rmentioning

confidence: 99%

“…Next section compares the 1D (13) and 3D (19) one-sparse GLR tests for the K-SVD [25] and the minimax approaches of [26].…”

Section: Design Strategies For Rmentioning

confidence: 99%

Section: Introduction and Previous Workmentioning

confidence: 99%

See 2 more Smart Citations

Constrained likelihood ratios for detecting sparse signals in highly noisy 3D data

Paris

Suleiman

Mary

et al. 2013

2013 IEEE International Conference on Acoustics, Speech and Signal Processing

Self Cite

View full text Add to dashboard Cite

show abstract

“…In [5], a constrained likelihood ratio was proposed that exploited the spread of the signal in three dimensions using the PSF. A specific dictionary that was adapted to the signal to be detected was used to decrease the false-alarm probability, and this dictionary was optimized by techniques such as kernel singular value decomposition [7] or minimax [8]. All of these methods perform pixel-wise processing and can lead to tremendous computational issues.…”

Section: Introductionmentioning

confidence: 99%

Non-parametric Bayesian framework for detection of object configurations with large intensity dynamics in highly noisy hyperspectral data

Meillier

Chatelain

Michel

et al. 2014

2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

View full text Add to dashboard Cite

To cite this version:Céline Meillier, Florent Chatelain, Olivier Michel, Hacheme Ayasso. Non-parametric Bayesian framework for detection of object configurations with large intensity dynamics in highly noisy hyperspectral data. IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2014), May 2014, Florence, Italy. Proceedings of ICASSP 2014, pp.1905-1909, 2014 ABSTRACTIn this study, a method that aims at detecting small and faint objects in noisy hyperspectral astrophysical images is presented. The particularity of the hyperspectral images that we are interested in is the high dynamics between object intensities. Detection of the smallest and faintest objects is challenging, because their signal-to-noise ratio is low, and if the brightest objects are not well reconstructed, their residuals can be more energetic than faint objects. This paper proposes a marked point process within a nonparametric Bayesian framework for the detection of galaxies in hyperspectral data. The efficiency of the method is demonstrated on synthetic images, and it provides good results for very faint objects in quasi-real astrophysical hyperspectral data.

show abstract

Subspace learning in minimax detection

Suleiman

Mary

Ferrari

2014

2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

View full text Add to dashboard Cite

We consider the problem where a large known library of L alternatives is available and we wish to maximize the detection power in a worst case scenario. The considered minimax detection approach relies on a GLR test allied to a sparsity constraint. This approach conditions the optimization of the target subspaces, in number r L. While the exact solution of the minimax optimization problem can be found for r = 1, the problem for r > 1 is more intricate and we propose two algorithms aimed at finding an approximate solution. The proposed algorithms are illustrated on a face database and on hyperspectral data and are shown to improve on the r = 1 case.

show abstract

Minimax sparse detection based on one-class classifiers

Cited by 7 publications

References 24 publications

Constrained likelihood ratios for detecting sparse signals in highly noisy 3D data

Constrained likelihood ratios for detecting sparse signals in highly noisy 3D data

Non-parametric Bayesian framework for detection of object configurations with large intensity dynamics in highly noisy hyperspectral data

Subspace learning in minimax detection

Contact Info

Product

Resources

About