Multi-View Oriented GPLVM: Expressiveness and Efficiency

Zi Yang (Jiangnan University) · Ying Li (University of Hong Kong) · Zhidi Lin (University of Hong Kong) · Michael Minyi Zhang (University of Hong Kong) · Pablo Martinez Olmos (Universidad Carlos III de Madrid)

bivariate gaussian mixturecomputational efficiency gaussian processkernel expressivenesslatent representations latent variable modelmeaningful representationsmulti-view datanext-gen spectral mixturenumerical evaluationsrandom fourier feature approximationreparameterization trickscalable variational inferencespectral densitystate-of-the-art models

The multi-view Gaussian process latent variable model (MV-GPLVM) aims to learn a unified representation from multi-view data but is hindered by challenges such as limited kernel expressiveness and low computational efficiency. To overcome these issues, we first introduce a new duality between the spectral density and the kernel function. By modeling the spectral density with a bivariate Gaussian mixture, we then derive a generic and expressive kernel termed Next-Gen Spectral Mixture (NG-SM) for MV-GPLVMs. To address the inherent computational inefficiency of the NG-SM kernel, we propose a random Fourier feature approximation. Combined with a tailored reparameterization trick, this approximation enables scalable variational inference for both the model and the unified latent representations. Numerical evaluations across a diverse range of multi-view datasets demonstrate that our proposed method consistently outperforms state-of-the-art models in learning meaningful latent representations.