噪声预训练对隐式神经表示的惊人有效性
The Surprising Effectiveness of Noise Pretraining for Implicit Neural Representations
用不同噪声类（均匀、高斯、Dead Leaves、1/|f|^α 频谱）预训练 INR 的初始化参数，可替代数据驱动的初始化方法；其中 1/|f|^α 频谱噪声预训练在信号拟合与去噪逆问题上取得最佳平衡，为缺乏领域数据的 INR 应用提供高效初始化方案。
原文依据 noise with the classic 1/|f^\alpha| spectral structure of natural images achieves an excellent balance of signal fitting and inverse imaging capabilities
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论隐式神经表示（INR）的近似与收敛性质已知对参数初始化策略高度敏感。尽管若干数据驱动的初始化方法相比标准随机采样展现出显著改进，但其成功原因——具体而言，是编码了经典统计信号先验还是更复杂的特征——仍鲜为人知。本研究通过一系列利用噪声预训练的实验分析来探索这一现象。我们在多种噪声类别（如高斯噪声、死叶噪声、频谱噪声）上预训练 INR，并测量其拟合未见信号以及为逆成像任务（去噪）编码先验的能力。我们在图像和视频数据上的分析揭示了一个令人惊讶的发现：仅对无结构噪声（均匀噪声、高斯噪声）进行预训练，就能相比所有其他基线大幅提升信号拟合能力。然而，无结构噪声对于去噪任务产生的深度图像先验效果较差。相反，我们还发现，具有自然图像经典 1/|f^α| 频谱结构的噪声在信号拟合与逆成像能力之间取得了极佳平衡，其表现与最佳数据驱动初始化方法相当。这一发现使得在缺乏足够先验领域数据的应用中能够更高效地训练 INR。
英文原文摘要 1391 字符
The approximation and convergence properties of implicit neural representations (INRs) are known to be highly sensitive to parameter initialization strategies. While several data-driven initialization methods demonstrate significant improvements over standard random sampling, the reasons for their success -- specifically, whether they encode classical statistical signal priors or more complex features -- remain poorly understood. In this study, we explore this phenomenon through a series of experimental analyses leveraging noise pretraining. We pretrain INRs on diverse noise classes (e.g., Gaussian, Dead Leaves, Spectral) and measure their ability to both fit unseen signals and encode priors for an inverse imaging task (denoising). Our analyses on image and video data reveal a surprising finding: simply pretraining on unstructured noise (Uniform, Gaussian) dramatically improves signal fitting capacity compared to all other baselines. However, unstructured noise also yields poor deep image priors for denoising. In contrast, we also find that noise with the classic 1/|f^\alpha| spectral structure of natural images achieves an excellent balance of signal fitting and inverse imaging capabilities, performing on par with the best data-driven initialization methods. This finding enables more efficient INR training in applications lacking sufficient prior domain-specific data.