Abstract:We introduce the concept of mixture inner product spaces associated with a given separable Hilbert space, which feature an infinite-dimensional mixture of finite-dimensional vector spaces and are dense in the underlying Hilbert space. Any Hilbert valued random element can be arbitrarily closely approximated by mixture inner product space valued random elements. While this concept can be applied to data in any infinite-dimensional Hilbert space, the case of functional data that are random elements in the L space of square integrable functions is of special interest. For functional data, mixture inner product spaces provide a new perspective, where each realization of the underlying stochastic process falls into one of the component spaces and is represented by a finite number of basis functions, the number of which corresponds to the dimension of the component space. In the mixture representation of functional data, the number of included mixture components used to represent a given random element in L is specifically adapted to each random trajectory and may be arbitrarily large. Key benefits of this novel approach are, first, that it provides a new perspective on the construction of a probability density in function space under mild regularity conditions, and second, that individual trajectories possess a trajectory-specific dimension that corresponds to a latent random variable, making it possible to use a larger number of components for less smooth and a smaller number for smoother trajectories. This enables flexible and parsimonious modeling of heterogeneous trajectory shapes. We establish estimation consistency of the functional mixture density and introduce an algorithm for fitting the functional mixture model based on a modified expectation-maximization algorithm. Simulations confirm that in comparison to traditional functional principal component analysis the proposed method achieves similar or better data recovery while using fewer components on average. Its practical merits are also demonstrated in an analysis of egg-laying trajectories for medflies.

Kernel Mean Embedding of Probability Measures and its Applications to Functional Data Analysis

Statistical Depth Meets Machine Learning: Kernel Mean Embeddings and Depth in Functional Data Analysis

Multivariate Statistical Process Monitoring Using Kernel Density Estimation

Keep it Tighter -- A Story on Analytical Mean Embeddings

Variance-Aware Estimation of Kernel Mean Embedding

Measures of Entropy From Data Using Infinitely Divisible Kernels

A Fourier representation of kernel Stein discrepancy with application to Goodness-of-Fit tests for measures on infinite dimensional Hilbert spaces

Diffeomorphic Measure Matching with Kernels for Generative Modeling

Learning to Embed Distributions via Maximum Kernel Entropy

FUSED KERNEL-SPLINE SMOOTHING FOR REPEATEDLY MEASURED OUTCOMES IN A GENERALIZED PARTIALLY LINEAR MODEL WITH FUNCTIONAL SINGLE INDEX

An Overview of Causal Inference using Kernel Embeddings

Depth based inference on conditional distribution with infinite dimensional data

The functional mean-shift algorithm for mode hunting and clustering in infinite dimensions

Kernel-based independence tests for causal structure learning on functional data

Semblance: A Rank-Based Kernel on Probability Spaces for Niche Detection

Roduct s paces and t heir a pplication to f unctional d ata a nalysis

Entropy Measures for Stochastic Processes with Applications in Functional Anomaly Detection

Ultra-Fast Approximate Inference Using Variational Functional Mixed Models

EMPIRICAL LIKELIHOOD‐BASED KERNEL DENSITY ESTIMATION

On kernel-based statistical learning in the mean field limit