Latent variable modeling with random features
Gregory W. Gundersen, Michael Minyi Zhang, Barbara E. Engelhardt · arXiv (Cornell University) · 2020
Gaussian process-based latent variable models are flexible and theoretically grounded tools for nonlinear dimension reduction, but generalizing to non-Gaussian data likelihoods within this nonlinear framework is statistically challenging. Here, we use random features to develop a family of nonlinear dimension reduction models that are easily extensible to non-Gaussian data likelihoods; we call these random feature latent variable models (RFLVMs). By approximating a nonlinear relationship between the latent space and the observations with a function that is linear with respect to random features, we induce closed-form gradients of the posterior distribution with respect to the latent variable. This allows the RFLVM framework to support computationally tractable nonlinear latent variable models for a variety of data likelihoods in the exponential family without specialized derivations. Our generalized RFLVMs produce results comparable with other state-of-the-art dimension reduction methods on diverse types of data, including neural spike train recordings, images, and text data.
Cited by 1 works. doi:10.48550/arxiv.2006.11145 Open-access PDF
Authors
References
- Bayesian Data Analysis
- A Global Geometric Framework for Nonlinear Dimensionality Reduction
- Learning the parts of objects by non-negative matrix factorization
- Gaussian Processes for Machine Learning
- Factor Analysis as a Statistical Method.
- A Bayesian Analysis of Some Nonparametric Problems
- Probabilistic Principal Component Analysis
- Some results on Tchebycheffian spline functions
- Mixtures of Dirichlet Processes with Applications to Bayesian Nonparametric Problems
- Bayesian Density Estimation and Inference Using Mixtures