IEEE

Low-Rank Matrix Recovery with Total Generalized Variation for Defending Adversarial Examples

Abstract: Low-rank matrix decomposition with first-order total variation (TV) regularization exhibits excellent performance in exploration of image structure. Taking advantage of its excellent ...
GitHub

A Python demonstration of speeding up matrix multiplication and reducing memory using a quantum-inspired low-rank SVD trick, with examples in AI.

This repository demonstrates a powerful, classical linear algebra technique—low-rank approximation via Singular Value Decomposition (SVD)—to dramatically accelerate common matrix operations like GEMM ...