Data Fusion by Matrix Factorization

By Marinka Zitnik, Blaz Zupan. 2013

209 citations recorded by the source corpus.

For most problems in science and engineering we can obtain data sets that describe the observed system from various perspectives and record the behavior of its individual components. Heterogeneous data sets can be collectively mined by data fusion. Fusion can focus on a specific target relation and exploit directly associated data together with contextual data and data about system's constraints. In the paper we describe a data fusion approach with penalized matrix tri-factorization (DFMF) that simultaneously factorizes data matrices to reveal hidden associations. The approach can directly consider any data that can be expressed in a matrix, including those from feature-based representations, ontologies, associations and networks. We demonstrate the utility of DFMF for gene function prediction task with eleven different data sources and for prediction of pharmacologic actions by fusing six data sources. Our data fusion algorithm compares favorably to alternative data integration approaches and achieves higher accuracy than can be obtained from any single data source alone.

Subfields: Molecular Biology, Representation & Self-Supervised Learning

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Molecular Biology

In Representation & Self-Supervised Learning

Citations in the corpus

Predicting multicellular function through multi-layer tissue networks (2017, background): “(1) First, current methods for cellular function prediction on networks (Mostafavi and Morris, 2009; Radivojac et al., 2013; Vidulin et al., 2016; Zitnik and Zupan, 2015) do not model networks with regards to patterns that span tissues, organs, and cellular systems.”
Modeling polypharmacy side effects with graph convolutional networks (2018, neutral): “may have similar interaction patterns (Gottlieb et al., 2012; Vilar et al., 2012; Huang et al., 2014b; Li et al., 2015; Zitnik and Zupan, 2015; Sun et al., 2015; Li et al., 2017).”
Machine Learning for Integrating Data in Biology and Medicine: Principles, Practice, and Opportunities (2018, methodology): “In intermediate integration, a model, such as multiple kernel learning [40, 41], collective matrix factorization [42, 43, 33] or deep neural network [44, 45], learns a joint representation of many datasets.”
Deep multi-view learning methods: A review (2021, background): “ative matrix factorization (NMF) [110], a specific form of matrix factorization, has received significant attention in various domains [111, 112, 113].”
Graph embedding on biomedical networks: methods, applications and evaluations (2019, unknown): “external information can also help build computational models for biomedical networks. For example, Zhang et al. (2018c) incorporate drug and disease features into MF to learn better representations. Zitnik and Zupan (2014) incorporate prior information (e.g. gene network) as a vector or a matrix to”
Disentangling Interactions in the Microbiome: A Network Perspective (2016, background): “Data fusion approaches have been successfully applied to predict gene function [48], mine disease–disease associations [49], to predict drug toxicity [50], and infer gene networks [51].”
A Survey of Multi-View Representation Learning (2016, unknown): “…by combining data matrices across views, where the consensus ensures the mutual agreement on multiple views of data and the complementarity exploits the distinctive information contained in different views for learning the complete representation which is useful for classification and retrieval.”
A review on machine learning principles for multi-view biological data integration (2016, methodology): “scikit-fusion [93] Data fusion based on DFMF github.”