用加法融合替代拼接，让去噪和分类更稳定高效
Multi-Task Learning with Additive U-Net for Image Denoising and Classification
- 用加法门控融合取代拼接，控制信息流并保持维度一致
- 在联合训练中实现与拼接相当的去噪效果，且训练更稳定
- 浅层侧重去噪，深层支持分类，隐含任务解耦适合多任务场景
本文研究U-Net架构中加法跳连接融合在图像去噪及以去噪为中心的多任务学习（MTL）中的应用。通过将传统的拼接跳连接替换为门控加法融合，提出的Additive U-Net（AddUNet）在限制捷径容量的同时，保持各深度层的特征维度固定。这种结构正则化有效控制编码器-解码器间的信息流动，提升联合优化的稳定性。在单任务去噪与联合去噪-分类设置下，AddUNet实现了具有竞争力的重建性能，并显著改善训练稳定性。在多任务学习中，学习到的跳连接权重呈现出系统性的任务感知分配：浅层跳连接更倾向于重建任务，深层特征则支持分类判别。值得注意的是，即使分类能力受限，重建性能仍保持稳健，表明加法融合可实现隐式任务解耦。这些发现说明，对跳连接施加简单约束即可作为有效的架构正则化手段，在不增加模型复杂度的前提下，实现稳定且可扩展的多任务学习。
原文摘要 · Abstract (English)
We investigate additive skip fusion in U-Net architectures for image denoising and denoising-centric multi-task learning (MTL). By replacing concatenative skips with gated additive fusion, the proposed Additive U-Net (AddUNet) constrains shortcut capacity while preserving fixed feature dimensionality across depth. This structural regularization induces controlled encoder-decoder information flow and stabilizes joint optimization. Across single-task denoising and joint denoising-classification settings, AddUNet achieves competitive reconstruction performance with improved training stability. In MTL, learned skip weights exhibit systematic task-aware redistribution: shallow skips favor reconstruction, while deeper features support discrimination. Notably, reconstruction remains robust even under limited classification capacity, indicating implicit task decoupling through additive fusion. These findings show that simple constraints on skip connections act as an effective architectural regularizer for stable and scalable multi-task learning without increasing model complexity.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。