Abstract:Heavy tails is a common feature of filtering distributions that results from the nonlinear dynamical and observation processes as well as the uncertainty from physical sensors. In these settings, the Kalman filter and its ensemble version - the ensemble Kalman filter (EnKF) - that have been designed under Gaussian assumptions result in degraded performance. t-distributions are a parametric family of distributions whose tail-heaviness is modulated by a degree of freedom $\nu$. Interestingly, Cauchy and Gaussian distributions correspond to the extreme cases of a t-distribution for $\nu = 1$ and $\nu = \infty$, respectively. Leveraging tools from measure transport (Spantini et al., SIAM Review, 2022), we present a generalization of the EnKF whose prior-to-posterior update leads to exact inference for t-distributions. We demonstrate that this filter is less sensitive to outlying synthetic observations generated by the observation model for small $\nu$. Moreover, it recovers the Kalman filter for $\nu = \infty$. For nonlinear state-space models with heavy-tailed noise, we propose an algorithm to estimate the prior-to-posterior update from samples of joint forecast distribution of the states and observations. We rely on a regularized expectation-maximization (EM) algorithm to estimate the mean, scale matrix, and degree of freedom of heavy-tailed \textit{t}-distributions from limited samples (Finegold and Drton, arXiv preprint, 2014). Leveraging the conditional independence of the joint forecast distribution, we regularize the scale matrix with an $l1$ sparsity-promoting penalization of the log-likelihood at each iteration of the EM algorithm. By sequentially estimating the degree of freedom at each analysis step, our filter can adapt its prior-to-posterior update to the tail-heaviness of the data. We demonstrate the benefits of this new ensemble filter on challenging filtering problems.

$\ell_1$ Adaptive Trend Filter via Fast Coordinate Descent

Fast and Flexible ADMM Algorithms for Trend Filtering

Fused $\ell_{1}$ Trend Filtering on Graphs

On the Spectral Properties of Matrices Associated with Trend Filters

Trend Filtering -- I. A Modern Statistical Tool for Time-Domain Astronomy and Astronomical Spectroscopy

A recursive adaptive lattice filters for impulsive signals

Trend Filtering for Functional Data

Wasserstein total variation filtering

Robust Enhanced Trend Filtering With Unknown Noise

Multivariate Trend Filtering for Lattice Data

Minmax Trend Filtering: A Locally Adaptive Nonparametric Regression Method via Pointwise Min Max Optimization

RobustTrend: A Huber Loss with a Combined First and Second Order Difference Regularization for Time Series Trend Filtering

Deep Filtering With Adaptive Learning Rates

Towards Dynamic Trend Filtering through Trend Point Detection with Reinforcement Learning

A new set of asymmetric filters for tracking the short-term trend in real-time

Real time estimation in local polynomial regression, with application to trend-cycle analysis

Risk Bounds for Quantile Additive Trend Filtering

An adaptive ensemble filter for heavy-tailed distributions: tuning-free inflation and localization

A Fast L1 Linear Estimator and Its Application on Predictive Deconvolution

Improved Predictive Deep Temporal Neural Networks with Trend Filtering

Trend of high dimensional time series estimation using low-rank matrix factorization: heuristics and numerical experiments via the TrendTM package