File Download
Links for fulltext
(May Require Subscription)
- Publisher Website: 10.1109/TPAMI.2020.3048727
- Scopus: eid_2-s2.0-85099094337
- PMID: 33385310
- WOS: WOS:000803117500030
- Find via
Supplementary
- Citations:
- Appears in Collections:
Article: PRIMAL-GMM: PaRametrIc MAnifold Learning of Gaussian Mixture Models
Title | PRIMAL-GMM: PaRametrIc MAnifold Learning of Gaussian Mixture Models |
---|---|
Authors | |
Keywords | Dimensionality Reduction and Manifold Learning Gaussian Mixture Models Interpretability Unsupervised Learning Probabilistic Models |
Issue Date | 2022 |
Publisher | Institute of Electrical and Electronics Engineers. The Journal's web site is located at https://ieeexplore.ieee.org/xpl/RecentIssue.jsp?punumber=34 |
Citation | IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022, v. 44 n. 6, p. 3197-3211 How to Cite? |
Abstract | We propose a ParametRIc MAnifold Learning (PRIMAL) algorithm for Gaussian Mixtures Models (GMM), assuming that GMMs lie on or near to a manifold of probability distributions that is generated from a low-dimensional hierarchical latent space through parametric mappings. Inspired by Principal Component Analysis (PCA), the generative processes for priors, means and covariance matrices are modeled by their respective latent space and parametric mapping. Then, the dependencies between latent spaces are captured by a hierarchical latent space by a linear or kernelized mapping. The function parameters and hierarchical latent space are learned by minimizing the reconstruction error between ground-truth GMMs and manifold-generated GMMs, measured by Kullback-Leibler Divergence (KLD). Variational approximation is employed to handle the intractable KLD between GMMs and a variational EM algorithm is derived to optimize the objective function. Experiments on synthetic data, flow cytometry analysis, eye-fixation analysis and topic models show that PRIMAL learns a continuous and interpretable manifold of GMM distributions and achieves a minimum reconstruction error. |
Persistent Identifier | http://hdl.handle.net/10722/298687 |
ISSN | 2023 Impact Factor: 20.8 2023 SCImago Journal Rankings: 6.158 |
ISI Accession Number ID |
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Liu, Z | - |
dc.contributor.author | Yu, L | - |
dc.contributor.author | Hsiao, JH | - |
dc.contributor.author | Chan, AB | - |
dc.date.accessioned | 2021-04-12T03:02:00Z | - |
dc.date.available | 2021-04-12T03:02:00Z | - |
dc.date.issued | 2022 | - |
dc.identifier.citation | IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022, v. 44 n. 6, p. 3197-3211 | - |
dc.identifier.issn | 0162-8828 | - |
dc.identifier.uri | http://hdl.handle.net/10722/298687 | - |
dc.description.abstract | We propose a ParametRIc MAnifold Learning (PRIMAL) algorithm for Gaussian Mixtures Models (GMM), assuming that GMMs lie on or near to a manifold of probability distributions that is generated from a low-dimensional hierarchical latent space through parametric mappings. Inspired by Principal Component Analysis (PCA), the generative processes for priors, means and covariance matrices are modeled by their respective latent space and parametric mapping. Then, the dependencies between latent spaces are captured by a hierarchical latent space by a linear or kernelized mapping. The function parameters and hierarchical latent space are learned by minimizing the reconstruction error between ground-truth GMMs and manifold-generated GMMs, measured by Kullback-Leibler Divergence (KLD). Variational approximation is employed to handle the intractable KLD between GMMs and a variational EM algorithm is derived to optimize the objective function. Experiments on synthetic data, flow cytometry analysis, eye-fixation analysis and topic models show that PRIMAL learns a continuous and interpretable manifold of GMM distributions and achieves a minimum reconstruction error. | - |
dc.language | eng | - |
dc.publisher | Institute of Electrical and Electronics Engineers. The Journal's web site is located at https://ieeexplore.ieee.org/xpl/RecentIssue.jsp?punumber=34 | - |
dc.relation.ispartof | IEEE Transactions on Pattern Analysis and Machine Intelligence | - |
dc.rights | IEEE Transactions on Pattern Analysis and Machine Intelligence. Copyright © Institute of Electrical and Electronics Engineers. | - |
dc.rights | ©2021 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. | - |
dc.subject | Dimensionality Reduction and Manifold Learning | - |
dc.subject | Gaussian Mixture Models | - |
dc.subject | Interpretability | - |
dc.subject | Unsupervised Learning | - |
dc.subject | Probabilistic Models | - |
dc.title | PRIMAL-GMM: PaRametrIc MAnifold Learning of Gaussian Mixture Models | - |
dc.type | Article | - |
dc.identifier.email | Hsiao, JH: jhsiao@hku.hk | - |
dc.identifier.authority | Hsiao, JH=rp00632 | - |
dc.description.nature | postprint | - |
dc.identifier.doi | 10.1109/TPAMI.2020.3048727 | - |
dc.identifier.pmid | 33385310 | - |
dc.identifier.scopus | eid_2-s2.0-85099094337 | - |
dc.identifier.hkuros | 322147 | - |
dc.identifier.volume | 44 | - |
dc.identifier.issue | 6 | - |
dc.identifier.spage | 3197 | - |
dc.identifier.epage | 3211 | - |
dc.identifier.isi | WOS:000803117500030 | - |
dc.publisher.place | United States | - |