DICOD: Distributed Convolutional Sparse Coding

TL;DR

DICOD is a distributed convolutional sparse coding algorithm that efficiently builds shift-invariant representations for long signals, leveraging local message passing and greedy updates to accelerate convergence and improve computational speed.

Contribution

The paper introduces DICOD, a novel distributed convolutional sparse coding algorithm with proven convergence and super-linear speed-up, enhancing efficiency over existing methods.

Findings

Super-linear speed-up with increased cores

Empirical evidence of acceleration over state-of-the-art methods

Convergence proof for the distributed algorithm

Abstract

In this paper, we introduce DICOD, a convolutional sparse coding algorithm which builds shift invariant representations for long signals. This algorithm is designed to run in a distributed setting, with local message passing, making it communication efficient. It is based on coordinate descent and uses locally greedy updates which accelerate the resolution compared to greedy coordinate selection. We prove the convergence of this algorithm and highlight its computational speed-up which is super-linear in the number of cores used. We also provide empirical evidence for the acceleration properties of our algorithm compared to state-of-the-art methods.