arXiv subjects

S. M. Plis

Publications and source records attributed to S. M. Plis.

3 recordsLinked to original sources

arXiv2019

Learnt dynamics generalizes across tasks, datasets, and populations

Differentiating multivariate dynamic signals is a difficult learning problem as the feature space may be large yet often only a few training examples are available. Traditional approaches to this problem either proceed from handcrafted features or require large datasets to combat the m >> n problem. In this paper, we show that the source of the problem---signal dynamics---can be used to our advantage and noticeably improve classification performance on a range of discrimination tasks when training data is scarce. We demonstrate that self-supervised pre-training guided by signal dynamics produces embedding that generalizes across tasks, datasets, data collection sites, and data distributions. We perform an extensive evaluation of this approach on a range of tasks including simulated data, keyword detection problem, and a range of functional neuroimaging data, where we show that a single embedding learnt on healthy subjects generalizes across a number of disorders, age groups, and datasets.

cs.CV↗

arXiv2006

A generalized spatiotemporal covariance model for stationary background in analysis of MEG data

Using a noise covariance model based on a single Kronecker product of spatial and temporal covariance in the spatiotemporal analysis of MEG data was demonstrated to provide improvement in the results over that of the commonly used diagonal noise covariance model. In this paper we present a model that is a generalization of all of the above models. It describes models based on a single Kronecker product of spatial and temporal covariance as well as more complicated multi-pair models together with any intermediate form expressed as a sum of Kronecker products of spatial component matrices of reduced rank and their corresponding temporal covariance matrices. The model provides a framework for controlling the tradeoff between the described complexity of the background and computational demand for the analysis using this model. Ways to estimate the value of the parameter controlling this tradeoff are also discussed

physics.med-ph↗

arXiv2006

Modeling spatiotemporal noise covariance for MEG/EEG source analysis

We propose a new model for approximating spatiotemporal noise covariance for use in MEG/EEG source analysis. Our model is an extension of an existing model [1,2] that uses a single Kronecker product of a pair of matrices - temporal and spatial covariance; we employ a series of Kronecker products in order to construct a better approximation of the full covariance. In contrast to the single-pair model that assumes the same temporal structure for all spatial components, the proposed model allows for distinct, independent time courses at each spatial component. This model better describes spatially and temporally correlated background activity. At the same time, inversion of the model is fast which makes it useful in the inverse analysis. We have explored two versions of the model. One is based on orthogonal spatial components of the background. The other, more general model, is based on independent spatial components. Performance of the new and previous models is compared in inverse solutions to a large number of single dipole problems with simulated time courses and background from authentic MEG data.

physics.data-an↗