AI Paper Atlas

Research map / Image generation / Image restoration

An Information-Theoretic Regularizer for Lossy Neural Image Compression

Yingwen Zhang, Meng Wang, Xihua Sheng, Peilin Chen, Junru Li, Li Zhang, Shiqi Wang

ICCV 2025 · Image generation › Image restoration · approach: Learned image compression · 0 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
add negative conditional source entropy regularizer to the rate-distortion lossMLIC++ · Natural images averageBD-rate: -3.03%Improved
add the regularizer to classic hyperprior, autoregressive and attention modelshyperprior, autoregressive, attention · Natural images averageBD-rate: -0.88%, -1.06%, -0.92%Improved
apply the regularizer to models evaluated on unseen domainsall five compression models · Out-of-domain (pixel, screen, game, pathology)BD-rate: -0.43% to -3.44%Improved
apply the regularizer on the attention model for pathology versus natural imagesattention · PathologyBD-rate: -0.92% -> -2.38%Improved
train longer to 4x10^6 steps with the regularizer (ablation)hyperprior · Kodak natural imagesBD-rate: -1.24% at 2.5M -> around 0.7% gain at 4MMixed
set regularization factor alpha too large or too smallhyperprior · Kodak natural imagesBD-rateHurt
add an H(U|X_hat) estimator to the regularizerregularizer · Kodak natural imagesBD-rateHurt
use a stronger autoregressive source entropy model for a hyperprior compression model (ablation)hyperprior · Kodak natural imagesBD-rate: average BD-Rate remains positiveHurt
use a weaker hyperprior source entropy model for an autoregressive compression model (ablation)autoregressive · Kodak natural imagesBD-rate: -0.20% at 1.5M stepsMixed
train an additional source entropy model during trainingMLIC++ · Training timerelative training time: 100% -> 129.8%Hurt

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