Sparse probabilistic projections

By Cédric Archambeau, Francis R. Bach. 2008

150 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Composing graphical models with neural networks for structured representations and fast inference (2016, methodology): “For the synthetic 1D dot video data, we trained an LDS SVAE on 80 random image sequences each of length 50, using one sequence per update, and show the model’s future predictions given a prefix of a longer sequence.”
Optimization with Sparsity-Inducing Penalties (2011, background): “An important computational (in particular for variational methods) and theoretical aspect of these prior distributions is that they can be expressed as scaled mixture of Gaussians [5, 30], and thus enter the classical framework of automatic relevance determination [85].”
Bayesian group factor analysis with structured sparsity (2016, methodology): “Extensions of this approach to multiple coupled observations y (1) i ∈ R p 1 ×1 , · · · , y (m) i ∈ R pm×1 have resulted in group factor analysis models (GFA) (Archambeau and Bach, 2009; Salzmann et al., 2010; Jia et al., 2010; Virtanen et al., 2012).”
Factorized Orthogonal Latent Spaces (2010, background): “…have tried to factorize the information and learn separate latent spaces for modeling the shared (i.e., correlated across the views) and private (i.e., independent between the views) components of the input signals (Archambeau and Bach, 2008; Klami and Kaski, 2008; Ek et al., 2008; Leen, 2008).”
A survey of multi-view machine learning (2013, background): “It was further extended to nonlinear subspace learning [3, 17] and sparse formulations [2, 10, 20].”
Spike and Slab Variational Inference for Multi-Task and Multiple Kernel Learning (2011, background): “This contrasts with previous approaches [3, 7, 8, 9, 13] in which soft-sparse priors that assign zero probability mass to the weights being exactly zero were used.”
Mean field variational Bayes for continuous sparse signal shrinkage: Pitfalls and remedies (2014, unknown): “Archambeau and Bach [2]; Armagan [3]; Wand, Ormerod, Padoan and Frürwirth [34]).”
A Survey of Multi-View Representation Learning (2016, unknown): “Through fully investigating the characteristics of the existing successful multi-view representation learning techniques, we observe that there are several common underlying principles behind their success: correlation, consensus, and complementarity principles.”
Mean field variational bayes for elaborate distributions (2011, methodology): “…have already enjoyed some use in MFVB. Examples include Tipping and Lawrence (2003) for t -based robust curve ﬁtting with ﬁxed degrees of freedom, Archambeau and Bach (2008) and Armagan (2009) for Laplace and other exponential power distributions and Girolami and Rogers (2006) and Consonni and…”
Bayesian Canonical correlation analysis (2013, methodology): “The IBFA model has recently been re-invented in the machine learning community by several authors (Klami and Kaski, 2006, 2008; Ek et al., 2008; Archambeau and Bach, 2009), resulting in probabilistic descriptions identical with that of Browne (1979).”
Multi-View Intact Space Learning (2015, background): “A number of works exploit the shared latent subspace across diverse views, such as canonical correlation analysis (CCA) [28], [29], its kernel extension [30], its probabilistic interpretation [31], and its sparse formulation [32].”
Structured sparsity through convex optimization (2011, methodology): “Alternative sparse methods Bayesian methods • Lasso: minimize ∑n i=1 (yi − wxi)(2) + λ‖w‖1 – Equivalent to MAP estimation with Gaussian likelihood and factorized Laplace prior p(w) ∝ ∏p j=1 e −λ|wj| (Seeger, 2008) – However, posterior puts zero weight on exact zeros • Heavy-tailed distributions as a”
Bayesian exponential family projections for coupled data sources (2010, methodology): “In the exponential family projection framework, ECCA is obtained by factoring the parameters as U = ( US U1 U2 ) and V = VS1 VS2V1 0 0 V2 , following the presentation of Archambeau and Bach (2009).”
Factorized Latent Spaces with Structured Sparsity (2010, background): “This has been shown for linear mappings [1, 11], as well as for non-linear ones [7, 14, 20].”
Sparse Bayesian Multi-Task Learning (2011, methodology): “Several of the multivariate equivalents have recently been used as priors to induce sparsity in the Bayesian paradigm, both in the context of supervised [19, 11] and unsupervised linear Gaussian models [4].”