Registration of Functional Data Using Fisher-Rao Metric

Abstract

We introduce a novel geometric framework for separating the phase and the amplitude variability in functional data of the type frequently studied in growth curve analysis. This framework uses the Fisher-Rao Riemannian metric to derive a proper distance on the quotient space of functions modulo the time-warping group. A convenient square-root velocity function (SRVF) representation transforms the Fisher-Rao metric into the standard $\ltwo$ metric, simplifying the computations. This distance is then used to define a Karcher mean template and warp the individual functions to align them with the Karcher mean template. The strength of this framework is demonstrated by deriving a consistent estimator of a signal observed under random warping, scaling, and vertical translation. These ideas are demonstrated using both simulated and real data from different application domains: the Berkeley growth study, handwritten signature curves, neuroscience spike trains, and gene expression signals. The proposed method is empirically shown to be be superior in performance to several recently published methods for functional alignment.

Revised paper. More focused on a subproblem and more theoretical results

Cited by in corpus (32)

• An elementary introduction to information geometry
• Functional data analysis for density functions by transformation to a Hilbert space
• Generative Models for Functional Data using Phase and Amplitude Separation
• Functional Data Analysis of Amplitude and Phase Variation
• Statistical analysis of trajectories on Riemannian manifolds: Bird migration, hurricane tracking and video surveillance
• Amplitude and phase variation of point processes
• A Geometric Approach to Visualization of Variability in Functional Data
• scikit-fda: A Python Package for Functional Data Analysis
• Elastic Functional Principal Component Regression
• Regression with Distance Matrices
• Elastic analysis of irregularly or sparsely sampled curves
• Regression Models Using Shapes of Functions as Predictors
• Confidence Tubes for Curves on SO(3) and Identification of Subject-Specific Gait Change after Kneeling
• Estimation of Spatial Deformation for Nonstationary Processes via Variogram Alignment
• Multimodal Bayesian Registration of Noisy Functions using Hamiltonian Monte Carlo
• Time-warped growth processes, with applications to the modeling of boom-bust cycles in house prices
• Learning Registered Point Processes from Idiosyncratic Observations
• Bayesian registration of functions and curves
• Unsupervised Functional Data Analysis via Nonlinear Dimension Reduction
• Registration for Incomplete Non-Gaussian Functional Data
• Low-Rank Representation over the Manifold of Curves
• Functional outlier detection for density-valued data with application to robustify distribution to distribution regression
• Variograms for spatial functional data with phase variation
• Representations, Metrics and Statistics For Shape Analysis of Elastic Graphs
• Simultaneous Registration and Clustering for Multi-dimensional Functional Data
• Diffeomorphic Transformations for Time Series Analysis: An Efficient Approach to Nonlinear Warping
• Tractable Clustering of Data on the Curve Manifold
• Trend and Variable-Phase Seasonality Estimation from Functional Data
• Adapted Variational Bayes for Functional Data Registration, Smoothing, and Prediction
• Model-based curve registration via stochastic approximation EM algorithm
• Model-based Statistical Depth with Applications to Functional Data
• Discrepancy geometry in approximate Bayesian inference: transport and risk perspectives