Structure-revealing data fusion

Abstract

BackgroundAnalysis of data from multiple sources has the potential to enhance knowledge discovery by capturing underlying structures, which are, otherwise, difficult to extract. Fusing data from multiple sources has already proved useful in many applications in social network analysis, signal processing and bioinformatics. However, data fusion is challenging since data from multiple sources are often (i) heterogeneous (i.e., in the form of higher-order tensors and matrices), (ii) incomplete, and (iii) have both shared and unshared components. In order to address these challenges, in this paper, we introduce a novel unsupervised data fusion model based on joint factorization of matrices and higher-order tensors.ResultsWhile the traditional formulation of coupled matrix and tensor factorizations modeling only shared factors fails to capture the underlying structures in the presence of both shared and unshared factors, the proposed data fusion model has the potential to automatically reveal shared and unshared components through modeling constraints. Using numerical experiments, we demonstrate the effectiveness of the proposed approach in terms of identifying shared and unshared components. Furthermore, we measure a set of mixtures with known chemical composition using both LC-MS (Liquid Chromatography - Mass Spectrometry) and NMR (Nuclear Magnetic Resonance) and demonstrate that the structure-revealing data fusion model can (i) successfully capture the chemicals in the mixtures and extract the relative concentrations of the chemicals accurately, (ii) provide promising results in terms of identifying shared and unshared chemicals, and (iii) reveal the relevant patterns in LC-MS by coupling with the diffusion NMR data.ConclusionsWe have proposed a structure-revealing data fusion model that can jointly analyze heterogeneous, incomplete data sets with shared and unshared components and demonstrated its promising performance as well as potential limitations on both simulated and real data.Electronic supplementary materialThe online version of this article (doi:10.1186/1471-2105-15-239) contains supplementary material, which is available to authorized users.

ACMTF for fusion of multi-modal neuroimaging data and identification of biomarkers

“…The CP model is unique under mild conditions [19] and the ACMTF inherits uniqueness properties from CP [20]. In the presence of both shared/unshared components, 1-norm penalties on the weights help to obtain unique solutions [15].…”

“…In order to fit the ACMTF model, we use ACMTF-OPT [15] from the CMTF Toolbox and the nonlinear conjugate gradient Parts of sr corresponding to the time samples for each electrode in EEG and voxels in fMRI and sMRI are plotted. In EEG plots, green dashed plots show signals from individual electrodes averaged across all subjects.…”

“…The CMTF model used in [3] assumes that all extracted factors are shared by EEG and fMRI data. Rather than such a CMTF model, which may fail to provide a unique solution in the presence of both shared/unshared components [15], we use a structure-revealing CMTF model, known as the advanced CMTF (ACMTF) model [15], revealing the weights of components in each modality to identify shared/unshared factors and understand the contribution from each modality. Previously, we have shown the promise of the ACMTF model in a joint analysis of fMRI and multi-channel EEG data in terms of capturing neural patterns that can differentiate patients with schizophrenia from healthy controls [16].…”

Common and distinct components in data fusion

“…From the patterns of these 0's, the common and distinct components are defined, and thus, this method also obeys Equation . This type of approach—albeit in fusing 3‐ and 2‐way data—has been used in metabolomics . Structure‐revealing data fusion is within and between block scale dependent.…”

“…This type of approach-albeit in fusing 3-and 2-way data-has been used in metabolomics. 7,16 Structure-revealing data fusion is within and between block scale dependent.…”