Coordinate ascent neural Kalman-MLE for state estimation

A new paper introduces a coordinate ascent algorithm designed to enhance state estimation through maximum likelihood estimation. This approach focuses on learning dynamic and measurement models, particularly under the assumption that they are Gaussian. By optimizing neural network parameters, the algorithm effectively models dynamic functions and noise covariance matrices. This advancement is significant as it could improve the accuracy of state estimation in various applications, making it a noteworthy contribution to the field.