AI Paper Atlas

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

Multi-Sample Training for Neural Image Compression

Tongda Xu, Yan Wang, Dailan He, Chenjian Gao, Han Gao, Kunzan Liu, Hongwei Qin

NeurIPS 2022 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Variational methods for sequence models · 1 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (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
train with multiple-sample MIX objective (sample size 8) instead of single-sampleMS-NIC-MIX · Kodak model APSNR BD-BR (%): -3.847Improved
train with multiple-sample DMS objective (sample size 16) instead of single-sampleMS-NIC-DMS · Kodak model APSNR BD-BR (%): -4.929Improved
apply MS-NIC to autoregressive-context modelMS-NIC-DMS · Kodak model BPSNR BD-BR (%): -2.378Improved
apply MS-NIC to autoregressive-context modelMS-NIC-MIX · Kodak model BMS-SSIM BD-BR (%): 2.238Hurt
use original IWAE objective without direct-y trickIWAE · Kodak model APSNR BD-BR (%): 64.23Hurt
use IWAE-DReG gradient instead of pathwise gradient (ablation)MS-NIC-MIX (5 samples) · Kodak lambda=0.015R-D cost: 1.003 -> 1.058Hurt
remove direct-y trick (feed noisy y~ to z inference) (ablation)2-level VAE · Kodak lambda=0.015R-D cost: 1.017 -> 1.493Hurt
increase batch size x3 to x16 instead of importance-weighted samples (ablation)Baseline · Kodak lambda=0.015R-D cost: 1.017 -> 1.013No clear effect
use MS-NIC-MIX sample size 8 vs 16 (ablation)MS-NIC-MIX · Kodak lambda=0.015R-D cost: 0.9988 -> 1.013Mixed
increase MS-NIC-DMS sample size 3,3 to 16,16 (ablation)MS-NIC-DMS · Kodak lambda=0.015R-D cost: 1.010 -> 0.9954Improved
increase IWAE sample size without direct-y (ablation)IWAE · Kodak lambda=0.015R-D cost: 1.400 -> 1.748Hurt
use universal quantization instead of deterministic rounding (ablation)Universal Quantization · Kodak image 1R-D cost: 0.7552 -> 0.9080Hurt

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