arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2410.05050cs.LGcs.AIstat.ML
收藏

FreSh:用于加速神经表征学习的频率偏移方法

FreSh: Frequency Shifting for Accelerated Neural Representation Learning

Jagiellonian University(雅盖隆大学);ETH Zurich(苏黎世联邦理工学院);Balgrist University Hospital(巴尔格里斯大学医院);IDEAS NCBR

Adam Kania, Marko Mihajlovic, Sergey Prokudin, Jacek Tabor, Przemysław Spurek

AI总结:

针对隐式神经表征中MLP低频偏置及现有嵌入/激活超参数需昂贵网格搜索的问题,提出FreSh频率偏移方法,通过对齐初始输出与目标信号频谱选超参数,以极低开销实现媲美大规模搜索的性能提升。

摘要:

隐式神经表征(Implicit Neural Representations, INRs)作为一种利用多层感知器(multilayer perceptrons, MLPs)连续表征图像、视频和3D形状等信号的强大方法,近年来受到广泛关注。然而,MLP存在低频偏置问题,限制了其准确捕捉高频细节的能力。现有方法通常通过引入高频输入嵌入或特殊激活层来解决这一局限。本研究表明,这些嵌入和激活的超参数配置通常仅在平均意义上表现良好,但对于特定的输入信号并非最优,因此需要通过成本高昂的网格搜索来确定最佳设置。我们的核心观察是:未训练模型输出的初始频谱与模型在给定目标信号上的最终性能高度相关。基于这一洞见,我们提出了频率偏移(frequency shifting,简称FreSh)方法,该方法通过选择嵌入超参数,使模型初始输出的频谱与目标信号的频谱对齐。我们证明,这种简单的初始化技术能够在多种神经表征方法和任务中提升性能,达到与大规模超参数搜索相当的效果,但与使用默认超参数训练单个模型相比,仅增加了极少的计算开销。

英文摘要:

Implicit Neural Representations (INRs) have recently gained attention as a powerful approach for continuously representing signals such as images, videos, and 3D shapes using multilayer perceptrons (MLPs). However, MLPs are known to exhibit a low-frequency bias, limiting their ability to capture high-frequency details accurately. This limitation is typically addressed by incorporating high-frequency input embeddings or specialized activation layers. In this work, we demonstrate that these embeddings and activations are often configured with hyperparameters that perform well on average but are suboptimal for specific input signals under consideration, necessitating a costly grid search to identify optimal settings. Our key observation is that the initial frequency spectrum of an untrained model's output correlates strongly with the model's eventual performance on a given target signal. Leveraging this insight, we propose frequency shifting (or FreSh), a method that selects embedding hyperparameters to align the frequency spectrum of the model's initial output with that of the target signal. We show that this simple initialization technique improves performance across various neural representation methods and tasks, achieving results comparable to extensive hyperparameter sweeps but with only marginal computational overhead compared to training a single model with default hyperparameters.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑