Gaussian Process Latent Variable Models for Visualisation of High Dimensional Data

By Neil D. Lawrence. 2003

888 citations recorded by the source corpus.

Subfields: Representation & Self-Supervised Learning, Probabilistic & Bayesian Methods

How researchers see this paper

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

In Representation & Self-Supervised Learning

In Probabilistic & Bayesian Methods

Citations in the corpus

On the challenges of learning with inference networks on sparse, high-dimensional data (2017, background): “The assumption of linearity has been relaxed in nonlinear factor analysis (NFA) (Gibson, 1960) and extended across a variety of domains such as economics (Jones, 2006), signal processing (Jutten & Karhunen, 2003), and machine learning (Valpola & Karhunen, 2002; Lawrence, 2003).”
Non-Linear Regression for Bag-of-Words Data via Gaussian Process Latent Variable Set Model (2015, methodology): “The GP-LVSM is related to but different from the Gaussian process latent variable model (GP-LVM), which is used for dimension reduction (Lawrence 2004) and matrix factorization (Lawrence and Urtasun 2009).”
Deep Factors for Forecasting (2019, methodology): “Inspired by GP-LVM structure (Lawrence, 2004), Deep GPs stacks GPs on top of latent variables, resulting in more expressive mappings.”
Scalable Variational Gaussian Process Classification (2014, background): “, 2012], latent variable models [Lawrence, 2004] and deep GPs [Damianou and Lawrence, 2013], as the variational objective could be optimized as part of such a model without fear of overfitting.”
Meta Reinforcement Learning with Latent Variable Gaussian Processes (2018, methodology): “ more training data than the modelbased approaches we build up on in the present work. Our ML-GP model resembles the GP latent variable model (GPLVM), which is typically used in unsupervised settings [23]. In the GPLVM, the GP is used to map a low-dimensional latent embedding to higherdimensional ob”
Surpassing Human-Level Face Verification Performance on LFW with GaussianFace (2014, methodology): “e GaussianFace model. 7.1. Comparisons with Other MTGP/GP Methods Since our model is based on GPs, it is natural to compare our model with four popular GP models: GPC [42], MTGP prediction [6], GPLVM [29], and DGPLVM [57]. For fair comparisons, all these models are trained on multiple source-domain ”
Machine learning of linear differential equations using Gaussian processes (2017, background): “Such models generalize latent variable models [24,25,50] using differential equations.”
Factored conditional restricted Boltzmann Machines for modeling motion style (2009, methodology): “Recently, models based on the Gaussian Process Latent Variable Model (Lawrence, 2004) have been successfully applied to separate content and style in human motion (Wang et al.”
Pattern Recognition (2015, positive): “, determining the locations and orientations of all cameras within a common reference coordinate frame) is complicated by this setup since conventional calibration methods such as [26] rely on jointly observed patterns or objects with known geometry.”
Learning control Lyapunov function to ensure stability of dynamical system-based robot reaching motions (2014, methodology): “Various advantages and disadvantages can also be observed when using other techniques such as Support Vector Regression (SVR) [17], Reservoir Computing (RC) [18], and Gaussian Process Latent Variable Model (GPLVM) [19].”
Gaussian Processes for Signal Strength-Based Location Estimation (2006, methodology): “GP regression has been used with great success in a variety of applications, including sensor networks [3], data visualization [ 9 ], and computer animation [12].”
Probabilistic latent variable models for distinguishing between cause and effect (2010, methodology): “Alternatively, one could first integrate over the causal mechanisms f , and then optimize over the noise values e, similar to what is usually done in GPLVMs [16].”
Robust and Efficient Transfer Learning with Hidden Parameter Markov Decision Processes (2017, methodology): “By virtue of the GP (Lawrence 2004),(Urtasun and Darrell 2007) , this latent embedding allows the HiP-MDP to infer across similar task instances and provide a better prediction of the currently observed system.”
Hybrid and ensemble-based soft computing techniques in bankruptcy prediction: a survey (2010, methodology): “GP-LVM is a very promising non-linear mapping technique.”
Gaussian Process Dynamical Models (2005, positive): “Our approach is most directly inspired by the unsupervised Gaussian Process Latent Variable Model (GPLVM) [5], which models the joint distribution of the observed data and their corresponding representation in a low dimensional latent space.”
Latent Doodle Space (2006, methodology): “The Gaussian Process Latent Variable Model (GPLVM), introduced by Lawrence [Law04], automatically ﬁnds optimal positions for the latent variables according to a Gaussian process likelihood model.”