Dynamic Parameter Scheduling in Soft-Hard BPGD for Lossy Source Coding

TL;DR

This paper proposes a dynamic parameter scheduling method for soft-hard BPGD in lossy source coding, improving performance and reducing tuning complexity by adapting parameters during decimation.

Contribution

It introduces a novel dynamic scheduling framework for parameters in soft-hard BPGD, replacing exhaustive empirical tuning with a progressive schedule during decoding.

Findings

Enhanced rate-distortion performance with dynamic scheduling.

Reduced non-convergence compared to fixed-parameter methods.

Eliminated the need for extensive grid searches for optimal parameters.

Abstract

We investigate lossy source coding based on a soft-decision belief propagation guided decimation (BPGD) encoder for low-density generator matrix (LDGM) codes, referred to as \emph{soft-hard BPGD}. The performance of this encoder is highly sensitive to the choice of ``softness'' parameters, typically denoted by $(\beta,\mu)$, which are conventionally tuned via exhaustive empirical sweeps. To reduce this burden and to better align the algorithm with the evolving graphical structure during decimation, we introduce a \emph{dynamic scheduling} framework in which $(\beta,\mu)$ are not fixed globally but change as decimation progresses. The schedule starts in a softer regime to encourage exploration and gradually hardens toward the end to promote convergence, similar to simulated annealing. We consider linear and exponential schedules, discuss their physical interpretation via an effective temperature viewpoint, and explain how they integrate with soft-hard BPGD without changing the order of magnitude of its complexity. Numerical experiments with irregular and semi-regular LDGM ensembles indicate improved rate-distortion performance and reduced non-convergence compared to constant-parameter baselines, while largely eliminating expensive grid searches for a single best pair $(\beta,\mu)$.