Research map / Image generation / Image restoration
Noise-Regularized Training for Learned Image Compression
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes training learned image compression models with explicit input-space noise injection as a regularizer, rather than only the latent uniform-noise proxy used today.
Problem
Learned image compression is trained almost exclusively on clean images, with the only stochastic perturbation being an additive uniform latent proxy used to keep the entropy model differentiable.
Approach
Adds noise in the input (image) domain during LIC training as a regularizer, motivated by the success of noise-based regularization (Tikhonov, contractive autoencoders, denoising score matching, diffusion) in other settings.
Experimental setup
Builds on: Tikhonov regularization denoising score matching contractive autoencoders
Closest papers by meaning
- Efficient Learned Image Compression without Entropy Coding (ICML 2026 · similarity 0.90)
- Towards Efficient Image Compression Without Autoregressive Models (NeurIPS 2023 · similarity 0.90)
- Balanced Rate-Distortion Optimization in Learned Image Compression (CVPR 2025 · similarity 0.88)
- Few-Shot Domain Adaptation for Learned Image Compression (AAAI 2025 · similarity 0.88)
- On Disentangled Training for Nonlinear Transform in Learned Image Compression (ICLR 2025 · similarity 0.87)
- Robustly overfitting latents for flexible neural image compression (NeurIPS 2024 · similarity 0.87)
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.