Publications
Journal Article
An AO-ADMM approach to constraining PARAFAC2 on all modes
SIAM Journal on Mathematics of Data Science (2022).Status: Accepted
An AO-ADMM approach to constraining PARAFAC2 on all modes
Analyzing multi-way measurements with variations across one mode of the dataset is a challenge in various fields including data mining, neuroscience and chemometrics. For example, measurements may evolve over time or have unaligned time profiles. The PARAFAC2 model has been successfully used to analyze such data by allowing the underlying factor matrices in one mode (i.e., the evolving mode) to change across slices. The traditional approach to fit a PARAFAC2 model is to use an alternating least squares-based algorithm, which handles the constant cross-product constraint of the PARAFAC2 model by implicitly estimating the evolving factor matrices. This approach makes imposing regularization on these factor matrices challenging. There is currently no algorithm to flexibly impose such regularization with general penalty functions and hard constraints. In order to address this challenge and to avoid the implicit estimation, in this paper, we propose an algorithm for fitting PARAFAC2 based on alternating optimization with the alternating direction method of multipliers (AO-ADMM). With numerical experiments on simulated data, we show that the proposed PARAFAC2 AO-ADMM approach allows for flexible constraints, recovers the underlying patterns accurately, and is computationally efficient compared to the state-of-the-art. We also apply our model to a real-world chromatography dataset, and show that constraining the evolving mode improves the interpretability of the extracted patterns.
Afilliation | Machine Learning |
Project(s) | Department of Data Science and Knowledge Discovery , TrACEr: Time-Aware ConstrainEd Multimodal Data Fusion |
Publication Type | Journal Article |
Year of Publication | 2022 |
Journal | SIAM Journal on Mathematics of Data Science |
Publisher | arXiv |
URL | https://arxiv.org/abs/2110.01278 |
Tracing Evolving Networks using Tensor Factorizations vs. ICA-based Approaches
Frontiers in Neuroscience 16 (2022).Status: Published
Tracing Evolving Networks using Tensor Factorizations vs. ICA-based Approaches
Afilliation | Machine Learning |
Project(s) | Department of Data Science and Knowledge Discovery , TrACEr: Time-Aware ConstrainEd Multimodal Data Fusion |
Publication Type | Journal Article |
Year of Publication | 2022 |
Journal | Frontiers in Neuroscience |
Volume | 16 |
Date Published | 04/2022 |
Publisher | Frontiers |
URL | https://www.frontiersin.org/article/10.3389/fnins.2022.861402 |
DOI | 10.3389/fnins.2022.861402 |
Talks, contributed
A Flexible Optimization Framework for Regularized Matrix-Tensor Factorizations with Linear Couplings
In Asilomar Conference on Signals, Systems, and Computers, 2021.Status: Published
A Flexible Optimization Framework for Regularized Matrix-Tensor Factorizations with Linear Couplings
An effective way of jointly analyzing data from multiple sources is through coupled matrix and tensor factorizations (CMTF). Different characteristics of datasets from multiple sources require to employ various regularizations, constraints, loss functions and different types of coupling structures between datasets. While existing algorithmic approaches for CMTF can incorporate constraints, linear couplings and different loss functions, none of them has been shown to achieve the flexibility to incorporate all. We propose a flexible algorithmic framework for coupled matrix and tensor factorizations, which utilizes Alternating Optimization (AO) and the Alternating Direction Method of Multipliers (ADMM). The framework facilitates the use of a variety of constraints, loss functions and couplings with linear transformations. Numerical experiments on simulated datasets and real data from chemometrics and hyperspectral super-resolution demonstrate that the proposed approach is accurate, flexible and computationally efficient with comparable or better performance than available CMTF algorithms.
While we focus on CANDECOMP/PARAFAC (CP) –based CMTF models, we will also briefly discuss the use of an AO-ADMM based algorithmic approach for fitting a PARAFAC2 model. We demonstrate that the proposed algorithmic approach enables imposing constraints in all modes, which has been a challenge using the traditional alternating least squares-based algorithm used for PARAFAC2.
Afilliation | Machine Learning |
Project(s) | Department of Data Science and Knowledge Discovery , TrACEr: Time-Aware ConstrainEd Multimodal Data Fusion |
Publication Type | Talks, contributed |
Year of Publication | 2021 |
Location of Talk | Asilomar Conference on Signals, Systems, and Computers |
Type of Talk | Invited Session Talk |
Tracing Dynamic Networks through Constrained Parafac2 Decomposition
In SIAM Conference on Applied Linear Algebra (LA21), Virtual Conference. SIAM, 2021.Status: Published
Tracing Dynamic Networks through Constrained Parafac2 Decomposition
Time-evolving data analysis is crucial for understanding complex systems such as the brain. Methods that assume static networks have successfully recovered spatial networks of connectivity from neuroimaging data. Still, discovering both underlying networks and their evolution is a challenging task.
To capture temporal evolution of connectivity networks, we arrange dynamic data as a tensor and use a tensor factorization method called PARAFAC2. PARAFAC2 deciphers the hidden structure of dynamic networks and yields unique and interpretable components. Preliminary results using PARAFAC2 in neuroimaging data analysis are promising. However, the constant cross-product constraint on the time-evolving mode hinders the use of additional constraints or regularization (e.g. spatial smoothness) on this mode. Currently, the only way to regularize the time-evolving mode of a PARAFAC2 model is with a flexible coupling approach, which finds the solution through regularized least-squares subproblems. Instead, we use an alternating direction method of multipliers (ADMM) based approach to widen the possible regularization penalties to any proximable function.
Our numerical experiments demonstrate that the proposed ADMM-based algorithmic approach for PARAFAC2 can accurately recover the underlying evolving components, is flexible in terms of imposing constraints and also computationally efficient.
Afilliation | Machine Learning |
Project(s) | Department of Data Science and Knowledge Discovery , TrACEr: Time-Aware ConstrainEd Multimodal Data Fusion |
Publication Type | Talks, contributed |
Year of Publication | 2021 |
Location of Talk | SIAM Conference on Applied Linear Algebra (LA21), Virtual Conference |
Publisher | SIAM |
Proceedings, refereed
PARAFAC2 AO-ADMM: Constraints in all modes
In 2021 29th European Signal Processing Conference (EUSIPCO). IEEE, 2021.Status: Published
PARAFAC2 AO-ADMM: Constraints in all modes
Afilliation | Machine Learning |
Project(s) | Department of Data Science and Knowledge Discovery , TrACEr: Time-Aware ConstrainEd Multimodal Data Fusion |
Publication Type | Proceedings, refereed |
Year of Publication | 2021 |
Conference Name | 2021 29th European Signal Processing Conference (EUSIPCO) |
Pagination | 1040-1044 |
Publisher | IEEE |
Poster
Understanding the Dynamics of Complex Systems through Time-Evolving Data Mining
SIAM International Conference on Data Mining, 2021.Status: Published
Understanding the Dynamics of Complex Systems through Time-Evolving Data Mining
Afilliation | Machine Learning |
Project(s) | Department of Data Science and Knowledge Discovery |
Publication Type | Poster |
Year of Publication | 2021 |
Place Published | SIAM International Conference on Data Mining |
Type of Work | Poster at SDM’21 Doctoral Forum |
Proceedings, refereed
Tracing Network Evolution Using The Parafac2 Model
In 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). Barcelona, Spain: IEEE, 2020.Status: Published
Tracing Network Evolution Using The Parafac2 Model
Afilliation | Machine Learning |
Project(s) | Department of Data Science and Knowledge Discovery |
Publication Type | Proceedings, refereed |
Year of Publication | 2020 |
Conference Name | 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) |
Publisher | IEEE |
Place Published | Barcelona, Spain |
URL | https://ieeexplore.ieee.org/xpl/mostRecentIssue.jsp?punumber=9040208http... |
DOI | 10.1109/ICASSP40776.202010.1109/ICASSP40776.2020.9053902 |