Progressive Compression with Universally Quantized Diffusion Models

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Yang, Yibo, Will, Justus C., Mandt, Stephan
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866916543175065600
author Yang, Yibo
Will, Justus C.
Mandt, Stephan
author_facet Yang, Yibo
Will, Justus C.
Mandt, Stephan
contents Diffusion probabilistic models have achieved mainstream success in many generative modeling tasks, from image generation to inverse problem solving. A distinct feature of these models is that they correspond to deep hierarchical latent variable models optimizing a variational evidence lower bound (ELBO) on the data likelihood. Drawing on a basic connection between likelihood modeling and compression, we explore the potential of diffusion models for progressive coding, resulting in a sequence of bits that can be incrementally transmitted and decoded with progressively improving reconstruction quality. Unlike prior work based on Gaussian diffusion or conditional diffusion models, we propose a new form of diffusion model with uniform noise in the forward process, whose negative ELBO corresponds to the end-to-end compression cost using universal quantization. We obtain promising first results on image compression, achieving competitive rate-distortion and rate-realism results on a wide range of bit-rates with a single model, bringing neural codecs a step closer to practical deployment.
format Preprint
id arxiv_https___arxiv_org_abs_2412_10935
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Progressive Compression with Universally Quantized Diffusion Models
Yang, Yibo
Will, Justus C.
Mandt, Stephan
Machine Learning
Computer Vision and Pattern Recognition
Diffusion probabilistic models have achieved mainstream success in many generative modeling tasks, from image generation to inverse problem solving. A distinct feature of these models is that they correspond to deep hierarchical latent variable models optimizing a variational evidence lower bound (ELBO) on the data likelihood. Drawing on a basic connection between likelihood modeling and compression, we explore the potential of diffusion models for progressive coding, resulting in a sequence of bits that can be incrementally transmitted and decoded with progressively improving reconstruction quality. Unlike prior work based on Gaussian diffusion or conditional diffusion models, we propose a new form of diffusion model with uniform noise in the forward process, whose negative ELBO corresponds to the end-to-end compression cost using universal quantization. We obtain promising first results on image compression, achieving competitive rate-distortion and rate-realism results on a wide range of bit-rates with a single model, bringing neural codecs a step closer to practical deployment.
title Progressive Compression with Universally Quantized Diffusion Models
topic Machine Learning
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2412.10935