HiLLoC: Lossless Image Compression with Hierarchical Latent Variable Models

Abstract

We make the following striking observation: fully convolutional VAE models trained on 32x32 ImageNet can generalize well, not just to 64x64 but also to far larger photographs, with no changes to the model. We use this property, applying fully convolutional models to lossless compression, demonstrating a method to scale the VAE-based 'Bits-Back with ANS' algorithm for lossless compression to large color photographs, and achieving state of the art for compression of full size ImageNet images. We release Craystack, an open source library for convenient prototyping of lossless compression using probabilistic models, along with full implementations of all of our compression results.

References in corpus (9)

• Auto-Encoding Variational Bayes
• BIVA: A Very Deep Hierarchy of Latent Variables for Generative Modeling
• Asymmetric numeral systems
• Soft Weight-Sharing for Neural Network Compression
• Generating High Fidelity Images with Subscale Pixel Networks and Multidimensional Upscaling
• Parallel Multiscale Autoregressive Density Estimation
• Compression with Flows via Local Bits-Back Coding
• Interleaved entropy coders
• Fast Generation for Convolutional Autoregressive Models

Cited by in corpus (5)

• Improving Inference for Neural Image Compression
• On the Out-of-distribution Generalization of Probabilistic Image Modelling
• IB-DRR: Incremental Learning with Information-Back Discrete Representation Replay
• Lossless Image Compression Using a Multi-Scale Progressive Statistical Model
• iVPF: Numerical Invertible Volume Preserving Flow for Efficient Lossless Compression