Robust feature selection for block covariance Bayesian models

pour, Ali Foroughi; Dalton, Lori A.

doi:10.1109/icassp.2017.7952646

Cited by 10 publications

(18 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Optimal Bayesian Filter (OBF) assumes all blocks have size one, i.e., all features are independent, and assumes the events {f ∈Ḡ} are independent a priori. In this case π * (f ) can be found in closed form with little computation cost [6,9]. OBF is optimal under its modeling assumptions.…”

Section: Optimal Bayesian Filtermentioning

confidence: 99%

“…Here an extended version of a synthetic model developed to mimic microarrays is used to generate data. The original model is introduced in [13], and has been extended in [8,9]. In these models features are markers or nonmarkers.…”

Section: Synthetic Microarray Simulationsmentioning

confidence: 99%

mentioning

confidence: 99%

“…While 2MNC-DGIB has outstanding performance when its assumptions are satisfied [7], it performs poorly when bad features are dependent [9]. On the other hand, CMNC-OBF and 2MNC-Robust have been shown to have robust performance across Bayesian models with blockdiagonal covariances [9]. CMNC-OBF is extremely fast and enjoys particularly excellent performance when markers are individually strong with low correlations, but, like all filter methods, may miss weak features that are of interest due to high correlations with strong features [6,9].…”

mentioning

confidence: 99%

“…On the other hand, CMNC-OBF and 2MNC-Robust have been shown to have robust performance across Bayesian models with blockdiagonal covariances [9]. CMNC-OBF is extremely fast and enjoys particularly excellent performance when markers are individually strong with low correlations, but, like all filter methods, may miss weak features that are of interest due to high correlations with strong features [6,9]. 2MNC-Robust is computationally very manageable and generally improves upon CMNC-OBF in the presence of correlations.…”

mentioning

confidence: 99%

See 4 more Smart Citations

Heuristic Algorithms for Feature Selection under Bayesian Models with Block-diagonal Covariance Structure

pour

Dalton

2017

Proceedings of the 8th ACM International Conference on Bioinformatics, Computational Biology,and Health Informatics

Self Cite

View full text Add to dashboard Cite

Background: Many bioinformatics studies aim to identify markers, or features, that can be used to discriminate between distinct groups. In problems where strong individual markers are not available, or where interactions between gene products are of primary interest, it may be necessary to consider combinations of features as a marker family. To this end, recent work proposes a hierarchical Bayesian framework for feature selection that places a prior on the set of features we wish to select and on the label-conditioned feature distribution. While an analytical posterior under Gaussian models with block covariance structures is available, the optimal feature selection algorithm for this model remains intractable since it requires evaluating the posterior over the space of all possible covariance block structures and feature-block assignments. To address this computational barrier, in prior work we proposed a simple suboptimal algorithm, 2MNC-Robust, with robust performance across the space of block structures. Here, we present three new heuristic feature selection algorithms. Results: The proposed algorithms outperform 2MNC-Robust and many other popular feature selection algorithms on synthetic data. In addition, enrichment analysis on real breast cancer, colon cancer, and Leukemia data indicates they also output many of the genes and pathways linked to the cancers under study. Conclusions: Bayesian feature selection is a promising framework for small-sample high-dimensional data, in particular biomarker discovery applications. When applied to cancer data these algorithms outputted many genes already shown to be involved in cancer as well as potentially new biomarkers. Furthermore, one of the proposed algorithms, SPM, outputs blocks of heavily correlated genes, particularly useful for studying gene interactions and gene networks.

show abstract

Section: Optimal Bayesian Filtermentioning

confidence: 99%

Section: Synthetic Microarray Simulationsmentioning

confidence: 99%

mentioning

confidence: 99%

mentioning

confidence: 99%

mentioning

confidence: 99%

See 3 more Smart Citations