发表机构
School of Fundamental Science; Engineering Waseda University Tokyo, Japan; School of Engineering Institute of Science Tokyo Tokyo, Japan（基础科学学校; 东京日本早稻田大学工程; 东京东京科学研究所工程学校）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
研究针对自回归上下文模型推理慢问题，提出受视频编码启发的无需训练的推理加速算法，通过重排推理顺序保持依赖关系，能加速预训练模型超13倍且维持率失真性能，还可权衡依赖关系实现更快解码。
AI 中文摘要
自回归上下文模型是学习图像压缩的基础，但存在串行推理速度慢的问题。现有加速方法如棋盘上下文需要架构更改和重新训练，不适用于预训练模型。我们受视频编码标准中的波前并行性启发，提出一种完全无需训练的推理时间加速算法。该方法将推理重新组织成最优的“交错”波前顺序，在保持自回归依赖关系的同时最小化顺序步骤。实验结果表明，我们的方法能在保持原始率失真性能的同时，将预训练自回归模型加速超13倍。还证明了通过权衡精确上下文依赖关系可实现更快解码。
英文摘要
Autoregressive context models are foundational for learned image compression,but they suffer from slow serial inference. Existing acceleration methods such as checkerboard context require architectural changes and retraining, thus are inapplicable to pre-trained models. We propose a completely training-free inference-time acceleration algorithm inspired by wavefront parallelism in video coding standards. Our method reorganizes inference into an optimal ``staggered'' wavefront order, minimizing sequential steps while maintaining exact autoregressive dependencies. Experimental results show our approach accelerates pre-trained autoregressive models (e.g., Cheng et al.) by more than $13\times$ while preserving the original rate-distortion performance. We also demonstrate that faster decoding is possible by trading off precise context dependencies. Source code will be available at https://github.com/tokkiwa/compressai-wavefront.
CommentsAccepted by MMSP 2026
