Advances in Data-Driven Methods for System Identification and Signal Processing

The field of system identification and signal processing is witnessing significant developments, driven by the increasing availability of data and the need for more accurate and efficient methods. Researchers are exploring innovative approaches to address challenges such as outlier detection, frequency response identification, and spectrum prediction. A key trend is the integration of advanced mathematical techniques, such as Riemannian geometry and fractional Fourier transforms, to improve the robustness and accuracy of these methods. Another area of focus is the development of data-driven approaches for uncertainty model identification and unfalsification, which can provide valuable insights into system behavior and performance. Notable papers in this area include: Leveraging the Christoffel Function for Outlier Detection in Data Streams, which introduces novel methods for outlier detection in data streams. Spectrum Prediction in the Fractional Fourier Domain with Adaptive Filtering, which proposes a framework for accurate spectrum prediction using adaptive fractional Fourier transforms. Riemannian Change Point Detection on Manifolds with Robust Centroid Estimation, which presents a method for change-point detection in streaming time series data using robust centroid estimation on Riemannian manifolds.

Sources

Leveraging the Christoffel Function for Outlier Detection in Data Streams

Frequency Response Identification of Low-Order Systems: Finite-Sample Analysis

Spectrum Prediction in the Fractional Fourier Domain with Adaptive Filtering

Riemannian Change Point Detection on Manifolds with Robust Centroid Estimation

Linear Trading Position with Sparse Spectrum

Closed-Form Input Design for Identification under Output Feedback with Perturbation Constraints

Set-membership identification of continuous-time MIMO systems via Tustin discretization

Optimistic vs Pessimistic Uncertainty Model Unfalsification

Fast numerical derivatives based on multi-interval Fourier extension

Built with on top of