Structured Pruning of Feedback Delay Networks

DOI10.48550/ARXIV.2610.38768MaRDI QIDQ7409129arXivFDO

Authors Facundo Franchino, Karolina Prawda

Publication date 28 October 2026

Published in arXiv

Tldr We present a structured pruning method for feedback delay networks that replaces the dense feedback matrix with stages of small orthogonal blocks and fixed permutations. This allows for significant arithmetic speedup and coefficient reduction over dense implementations while preserving spectral and temporal fidelity.

Full work available at URL https://doi.org/10.48550/ARXIV.2610.38768

arXiv classification

eess.ASeess.SPmath.NA

Cites work

• A feasible method for optimization with orthogonality constraints
• Adam: A Method for Stochastic Optimization
• An Adaptation of the Fast Fourier Transform for Parallel Processing
• Computation of Plain Unitary Rotations Transforming a General Matrix to Triangular Form
• Five-diagonal matrices and zeros of orthogonal polynomials on the unit circle
• Matrix Computations
• Modal Decomposition of Feedback Delay Networks
• Numerical Linear Algebra
• On Lossless Feedback Delay Networks
• PyTorch: An Imperative Style, High-Performance Deep Learning Library
• The Geometry of Algorithms with Orthogonality Constraints
• The mathematics of perfect shuffles
• Unitary Triangularization of a Nonsymmetric Matrix

This page was built for publication: Structured Pruning of Feedback Delay Networks

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7409129)