atlas

Research map / Probabilistic and scientific modeling / Variational and generative sampling

iFlow: Numerically Invertible Flows for Efficient Lossless Compression via a Uniform Coder

NeurIPS 2021 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Expressive invertible architectures · 10 citations (OpenAlex)

Read the paperCheck your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace rANS coder (LBB) with UBCS uniform coder in flow compressionFlow++ on CIFAR10 · CIFAR10 encodingcoding time (ms): 116 -> 21.0Helped
replace LBB local bits-back numerical-error coding with MST + UBCSFlow++ · CIFAR10bits per dimension: 3.118 -> 3.118No clear effect
replace iVPF's own coding with iFlow MST + UBCS on the iVPF architectureiVPF architecture · CIFAR10bits per dimension: 3.201 -> 3.196Helped
replace LBB with iFlowFlow++ · CIFAR10auxiliary bits: 39.86 -> 34.28Helped
replace iVPF coding with iFlowiVPF architecture · CIFAR10auxiliary bits: 6.00 -> 7.00Hurt
replace rANS with UBCS entropy coderentropy coder · coding bandwidth, 1 thread encoderM symbol/s: 5.1 -> 380Helped
replace iVPF with iFlow when testing on ImageNet32-trained modelgeneralization · CIFAR10bits per dimension: 3.49 -> 3.36Helped

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy