Bit-Swap: Recursive Bits-Back Coding for Lossless Compression with Hierarchical Latent Variables

TL;DR

Bit-Swap introduces a novel recursive bits-back coding scheme that efficiently compresses hierarchical latent variable models, outperforming existing methods in lossless compression rates.

Contribution

We propose Bit-Swap, a new hierarchical compression scheme that generalizes BB-ANS and achieves better rates for models with Markov chain structures.

Findings

Bit-Swap outperforms existing techniques in lossless compression.

Empirical results confirm superior compression rates.

Implementation available at GitHub.

Abstract

The bits-back argument suggests that latent variable models can be turned into lossless compression schemes. Translating the bits-back argument into efficient and practical lossless compression schemes for general latent variable models, however, is still an open problem. Bits-Back with Asymmetric Numeral Systems (BB-ANS), recently proposed by Townsend et al. (2019), makes bits-back coding practically feasible for latent variable models with one latent layer, but it is inefficient for hierarchical latent variable models. In this paper we propose Bit-Swap, a new compression scheme that generalizes BB-ANS and achieves strictly better compression rates for hierarchical latent variable models with Markov chain structure. Through experiments we verify that Bit-Swap results in lossless compression rates that are empirically superior to existing techniques. Our implementation is available at https://github.com/fhkingma/bitswap.