Tracing Network Evolution Using The Parafac2 Model

@article{Roald2020TracingNE,
  title={Tracing Network Evolution Using The Parafac2 Model},
  author={Marie Roald and Suchita Bhinge and Chunying Jia and Vince D. Calhoun and T. Adalı and Evrim Acar},
  journal={ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
  year={2020},
  pages={1100-1104}
}
  • M. Roald, Suchita Bhinge, E. Acar
  • Published 23 October 2019
  • Computer Science
  • ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
Characterizing time-evolving networks is a challenging task, but it is crucial for understanding the dynamic behavior of complex systems such as the brain. For instance, how spatial networks of functional connectivity in the brain evolve during a task is not well-understood. A traditional approach in neuroimaging data analysis is to make simplifications through the assumption of static spatial networks. In this paper, without assuming static networks in time and/or space, we arrange the… 

Figures and Tables from this paper

PARAFAC2 AO-ADMM: Constraints in all modes
TLDR
The proposed alternating direction method of multipliers (ADMM)-based algorithm for fitting PARAFAC2 can accurately recover the underlying components from simulated data while being both computationally efficient and flexible in terms of imposing constraints.
LogPar: Logistic PARAFAC2 Factorization for Temporal Binary Data with Missing Values
TLDR
This paper proposes Logistic PARAFAC2 (LogPar) by modeling the binary irregular Tensor with Bernoulli distribution parameterized by an underlying real-valued tensor and approximate the underlying tensor with a positive-unlabeled learning loss function to account for the missing values.
Tracing Evolving Networks Using Tensor Factorizations vs. ICA-Based Approaches
TLDR
This article arranges time-evolving fMRI data as a subjects by voxels by time windows tensor, and analyzes the tensor using a tensor factorization-based approach called a PARAFAC2 model to reveal spatial dynamics, showing that a component indicating statistically significant group difference between patients with schizophrenia and healthy controls is captured.
An AO-ADMM approach to constraining PARAFAC2 on all modes
TLDR
An algorithm for fitting PARAFAC2 based on alternating optimization with the alternating direction method of multipliers (AO-ADMM) is proposed and it is shown that constraining the evolving mode improves the interpretability of the extracted patterns.
Exploring Dynamic Metabolomics Data With Multiway Data Analysis: a Simulation Study
TLDR
Numerical experiments demonstrate that despite the increasing complexity of the studied dynamic metabolic models, tensor factorization methods CANDECOMP/PARAFAC(CP) and Parallel Profiles with Linear Dependences (Paralind) can disentangle the sources of variations and thereby reveal the underlying mechanisms and their dynamics.

References

SHOWING 1-10 OF 26 REFERENCES
Quantifying functional connectivity in multi‐subject fMRI data using component models
TLDR
A flexible cross‐validation approach that assesses the ability of the models to predict voxel‐wise covariance in new data, using three different measures of generalization was proposed, and reveals important differences between subspace models in terms of predictability, robustness, characterization of subject variability, and interpretability of the model parameters.
Modeling latency and shape changes in trial based neuroimaging data
TLDR
This work extends multi-linear decomposition to account for general temporal modeling within a convolutional representation and demonstrates how this alleviates degeneracy and helps to extract physiologically plausible components.
Temporally Factorized Network Modeling for Evolutionary Network Analysis
TLDR
This work shows that it is indeed possible to achieve the goal of representing the edge structure of the network purely as a function of time with the use of a matrix factorization, in which the entries are parameterized by time.
Link Prediction on Evolving Data Using Matrix and Tensor Factorizations
TLDR
This paper looks at bipartite graphs changing over time and considers matrix- and tensor-based methods for predicting links, and presents a weight-based method for collapsing multi-year data into a single matrix.
COPA: Constrained PARAFAC2 for Sparse & Large Datasets
TLDR
A COnstrained PARAFAC2 (COPA) method is proposed, which carefully incorporates optimization constraints such as temporal smoothness, sparsity, and non-negativity in the resulting factors and outperforms all the baselines attempting to handle a subset of the constraints in terms of speed and accuracy.
SDREGION: Fast Spotting of Changing Communities in Biological Networks
TLDR
A novel algorithm is designed, SDREGION, that identifies subgraphs that decrease or increase in density monotonically over time, referred to as d-region or i-regions, respectively, and is scalable with a time complexity of O(mlogn + nlogn) where m is the number of edges, and n is thenumber of vertices in a given dynamic graph.
Dynamic graph summarization: a tensor decomposition approach
TLDR
This research addresses the problem using tensor decomposition, which enables it to capture the multi-way structure of the time-evolving network, which is unique and is impossible to obtain with other approaches such as matrix factorization.
An improved approach to detection of amplitude of low-frequency fluctuation (ALFF) for resting-state fMRI: Fractional ALFF
Nonnegative PARAFAC2: A Flexible Coupling Approach
TLDR
In the following manuscript, a relaxation of the PARAFAC2 model is introduced, that allows for imposing nonnegativity constraints on the varying mode and its performance is studied on both synthetic and chemometrics data.
...
1
2
3
...