Mixtures of Gaussian Processes

By Volker Tresp. 2000

275 citations recorded by the source corpus.

Subfields: 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 Probabilistic & Bayesian Methods

Citations in the corpus

Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer (2017, background): “obs, 1994), the mixture-of-experts approach has been the subject of much research. Different types of expert architectures hae been proposed such as SVMs (Collobert et al., 2002), Gaussian Processes (Tresp, 2001; Theis & Bethge, 2015; Deisenroth & Ng, 2015), Dirichlet Processes (Shahbaba &am”
Near-optimal sensor placements in Gaussian processes (2005, unknown): “More generally, nonstationary behavior can be modeled using mixtures of Gaussian processes, in which each component has a different kernel function (Tresp, 2001).”
Anomaly Detection and Removal Using Non-Stationary Gaussian Processes (2015, methodology): “I. INTRODUCTION Gaussian processes (GPs) are experiencing a resurgence of interest.”
Twenty Years of Mixture of Experts (2012, methodology): “The gate can be a softmax of GPs [55], or it can follow a Dirichlet distribution [56], a DP [18], or be a Gaussian mixture model [7].”
Nonmyopic active learning of Gaussian processes: an exploration-exploitation approach (2007, methodology): “, Tresp (2000)), where the marginal distributions are mixtures of Gaussians, our construction uses a linear combination of GPs, hence, for fixed parameters, our nonstationary model is still a GP.”
Bayesian optimisation for Intelligent Environmental Monitoring (2012, methodology): “Other strategies for learning spatial models have been studied, such as Gaussian Process Mixtures [7, 9], SpaceTime Process Convolutions [10, 11] or Kernel DM+V/W algorithm [12].”
Learning to Compose Topic-Aware Mixture of Experts for Zero-Shot Video Captioning (2018, methodology): “Then MoE has been applied to various machine learning algorithms (Jordan and Jacobs 1994; Collobert, Bengio, and Bengio 2003), such as SVMs (Collobert, Bengio, and Bengio 2002), Gaussian Processes (Tresp 2001), and deep networks (Ahmed, Baig, and Torresani 2016; Gu et al. 2018).”
Multi-modal estimation with kernel embeddings for learning motion models (2013, background): “GP mixture models relax the Gaussianity assumption of the posterior, but retain a strong assumption of mixture component independence [25].”
Gaussian Processes for Regression and Optimisation (2007, methodology): “Mixtures of Gaussian processes were introduced by [81] followed by an extension to a tractable infinite mixture of Gaussian processes experts [59].”
Conditional Density Estimation via Least-Squares Density Ratio Estimation (2010, background): “Similarly, a mixture of Gaussian processes was explored for estimating the conditional density (Tresp, 2001).”
Near-Optimal Sensor Placements in Gaussian Processes: Theory, Efficient Algorithms and Empirical Studies (2008, unknown): “More generally, nonstationary behavior can be modeled using mixtures of Gaussian processes, in which each component has a different kernel function (Tresp, 2001).”
Gaussian Processes for Object Categorization (2010, methodology): “GPs have received limited attention in the computer vision literature to date perhaps due to the fact that they are conventionally limited to modest amounts of training data: the learning complexity is O(n 3 ) , cubic in the number of training examples.”
Parameter space exploration with Gaussian process trees (2004, methodology): “Alternative approaches for nonstationary modeling include mixtures of GPs (Tresp, 2001) and infinite mixtures of GPs (Rasmussen & Ghahramani, 2002).”
When Gaussian Process Meets Big Data: A Review of Scalable GPs (2018, background): “In 2001, Tresp [129] ﬁrst introduced the mixture of GP experts, which employs 3 M GP experts to respectively capture the mean, the noise variance, and the gate parameters with nearly O (3 Mn 3 ) complexity, which is unattractive for big data.”
When Training and Test Sets are Different: Characterising Learning Transfer (2013, background): “See [28] for more details of this approach to Gaussian process mixtures.”