Abstract:Practical Bayes filters often assume the state distribution of each time step to be Gaussian for computational tractability, resulting in the so-called Gaussian filters. When facing nonlinear systems, Gaussian filters such as extended Kalman filter (EKF) or unscented Kalman filter (UKF) typically rely on certain linearization techniques, which can introduce large estimation errors. To address this issue, this paper reconstructs the prediction and update steps of Gaussian filtering as solutions to two distinct optimization problems, whose optimal conditions are found to have analytical forms from Stein's lemma. It is observed that the stationary point for the prediction step requires calculating the first two moments of the prior distribution, which is equivalent to that step in existing moment-matching filters. In the update step, instead of linearizing the model to approximate the stationary points, we propose an iterative approach to directly minimize the update step's objective to avoid linearization errors. For the purpose of performing the steepest descent on the Gaussian manifold, we derive its natural gradient that leverages Fisher information matrix to adjust the gradient direction, accounting for the curvature of the parameter space. Combining this update step with moment matching in the prediction step, we introduce a new iterative filter for nonlinear systems called \textit{N}atural Gr\textit{a}dient Gaussia\textit{n} Appr\textit{o}ximation filter, or NANO filter for short. We prove that NANO filter locally converges to the optimal Gaussian approximation at each time step. Furthermore, the estimation error is proven exponentially bounded for nearly linear measurement equation and low noise levels through constructing a supermartingale-like property across consecutive time steps.

Selective Partial-Update Widely Linear Geometric-Algebra NLMS

Variable Partial-Update NLMS Algorithms with Data-Selective Updating.

Novel Improvement for Nonlinear Compatibility of Least Mean Square Adaptive Algorithm

Research on Robustness of Geometric Algebra Based Adaptive Filtering Algorithms In Non-Gaussian Environment

Geometry Transformation-Based Adaptive In-Loop Filter

Variants of Partial Update Augmented CLMS Algorithm and Their Performance Analysis

Non-Convex Projected Nlms Algorithm for Adaptive Nonlinear Prediction

Insights Into Algorithms for Separable Nonlinear Least Squares Problems

Nonlinear Multiple-Delay Feedback Based Kernel Least Mean Square Algorithm

Nonlinear Bayesian Filtering with Natural Gradient Gaussian Approximation

Efficient Adaptive Bilinear Filters for Nonlinear Active Noise Control

Improving S-MSCKF With Variational Bayesian Adaptive Nonlinear Filter

Robust Widely Nonlinear Quaternion Volterra Adaptive Filter Based On Recursive General Barron Algorithm and Its Performance Analysis

Understanding Several Adaptive Filter Algorithms Based on the Weight-update Strategy

Normalized LMS algorithm and data-selective strategies for adaptive graph signal estimation

Efficient subspace-based algorithm for adaptive direction estimation and tracking of narrowband signals

Lms and Frls 2-D Lattice Filters

Study of Set-Membership Adaptive Kernel Algorithms

Sparsity-Aware Adaptive Algorithms Based on Alternating Optimization with Shrinkage

Variable Step-Size NLMP Algorithm with a Gradient-Based Weighted Average in Impulsive Environments

An Adaptive -GLMB Filter Under Noise Statistics Mismatch