← 研踪 PaperTrailarXiv:2611.13707 · v1
Joint Activation and Weight Quantization for Learned Image Compression
首次提交 2026-11-10 16:00
版本更新 2026-11-10 16:00
雷达观测 2026-11-11 11:04
摘要
Learned image compression (LIC) models achieve competitive rate-distortion performance with increasingly large networks, making network quantization important for efficient deployment. Directly quantizing LIC models remains difficult in post-training quantization (PTQ) because compression performance is highly sensitive to quantization noise. Existing methods typically focus on optimizing either the weights or the activations, resulting in substantial rate-distortion degradation. To address this limitation, we introduce a two-stage pipeline that uses SmoothQuant to balance activation and weight ranges, followed by GPTQ to adapt the transformed weights. The proposed joint search framework optimizes the balancing transformation and adapted weights under the rate-distortion objective, accounting for the coupling between the two stages. Experiments across multiple LIC architectures and activation quantization configurations show that the joint search strategy consistently reduces rate-distortion degradation after quantization.
作者备注：Accepted at IEEE VCIP 2026. 5 pages, 3 figures
观测到的版本
相邻记录可能跳过中间版本；首次发现 v2 及以上也会标为版本更新。
v1 · 2026-11-10 16:00 · 最新观测版本
Joint Activation and Weight Quantization for Learned Image Compression
Learned image compression (LIC) models achieve competitive rate-distortion performance with increasingly large networks, making network quantization important for efficient deployment. Directly quantizing LIC models remains difficult in post-training quantization (PTQ) because compression performance is highly sensitive to quantization noise. Existing methods typically focus on optimizing either the weights or the activations, resulting in substantial rate-distortion degradation. To address this limitation, we introduce a two-stage pipeline that uses SmoothQuant to balance activation and weight ranges, followed by GPTQ to adapt the transformed weights. The proposed joint search framework optimizes the balancing transformation and adapted weights under the rate-distortion objective, accounting for the coupling between the two stages. Experiments across multiple LIC architectures and activation quantization configurations show that the joint search strategy consistently reduces rate-distortion degradation after quantization.
查看此版本解读 →查看此版本原文 ↗