18 December 2014 in IEEE Transactions on Neural Networks and Learning Systems
Factor analysis (FA) provides linear factors that describe the relationships between individual variables of a data set. We extend this classical formulation into linear factors that describe the relationships between groups of variables, where each group represents either a set of related variables or a data set. The model also naturally extends canonical correlation analysis to more than two sets, in a way that is more flexible than previous extensions. Our solution is formulated as a variational inference of a latent variable model with structural sparsity, and it consists of two hierarchical levels: 1) the higher level models the relationships between the groups and 2) the lower models the observed variables given the higher level. We show that the resulting solution solves the group factor analysis (GFA) problem accurately, outperforming alternative FA-based solutions as well as more straightforward implementations of GFA. The method is demonstrated on two life science data sets, one on brain activation and the other on systems biology, illustrating its applicability to the analysis of different types of high-dimensional data sources.
19 April 2016 in Bioinformatics
Motivation: Modelling methods that find structure in data are necessary with the current large volumes of genomic data, and there have been various efforts to find subsets of genes exhibiting consistent patterns over subsets of treatments. These biclustering techniques have focused on one data source, often gene expression data. We present a Bayesian approach for joint biclustering of multiple data sources, extending a recent method Group Factor Analysis (GFA) to have a biclustering interpretation with additional sparsity assumptions. The resulting method enables data-driven detection of linear structure present in parts of the data sources.
10 June 2016 in Machine Learning
1 January 2019 in Human Brain Mapping
Brain structure and many brain functions are known to be genetically controlled, but direct links between neuroimaging measures and their underlying cellular‐level determinants remain largely undiscovered. Here, we adopt a novel computational method for examining potential similarities in high‐dimensional brain imaging data between siblings. We examine oscillatory brain activity measured with magnetoencephalography (MEG) in 201 healthy siblings and apply Bayesian reduced‐rank regression to extract a low‐dimensional representation of familial features in the participants' spectral power structure. Our results show that the structure of the overall spectral power at 1–90 Hz is a highly conspicuous feature that not only relates siblings to each other but also has very high consistency within participants' own data, irrespective of the exact experimental state of the participant. The analysis is extended by seeking genetic associations for low‐dimensional descriptions of the oscillatory brain activity. The observed variability in the MEG spectral power structure was associated with SDK1 (sidekick cell adhesion molecule 1) and suggestively with several other genes that function, for example, in brain development. The current results highlight the potential of sophisticated computational methods in combining molecular and neuroimaging levels for exploring brain functions, even for high‐dimensional data limited to a few hundred participants.
25 March 2019 in Machine Learning
Bayesian matrix factorization (BMF) is a powerful tool for producing low-rank representations of matrices and for predicting missing values and providing confidence intervals. Scaling up the posterior inference for massive-scale matrices is challenging and requires distributing both data and computation over many workers, making communication the main computational bottleneck. Embarrassingly parallel inference would remove the communication needed, by using completely independent computations on different data subsets, but it suffers from the inherent unidentifiability of BMF solutions. We introduce a hierarchical decomposition of the joint posterior distribution, which couples the subset inferences, allowing for embarrassingly parallel computations in a sequence of at most three stages. Using an efficient approximate implementation, we show improvements empirically on both real and simulated data. Our distributed approach is able to achieve a speed-up of almost an order of magnitude over the full posterior, with a negligible effect on predictive accuracy. Our method outperforms state-of-the-art embarrassingly parallel MCMC methods in accuracy, and achieves results competitive to other available distributed and parallel implementations of BMF.