Abstract:Implicit Neural Representations (INRs) model continuous signals using multilayer perceptrons (MLPs), enabling compact, differentiable, and high-fidelity representations of data across diverse domains. However, due to the low-frequency bias of MLPs that prevents effective learning of small details, the model's frequency must be carefully tuned through the embedding layer. Prior work established that this tuning can be performed before training based on the target signal, but it did not account for the significant effect of model depth, indicating that our understanding of the relationship between frequency and INR performance remains limited. To gain insights into this relationship, we utilize the Spectral Energy Centroid (SEC) metric that quantifies the frequency of target images and the spectral bias of INR models. We show that SEC is a versatile tool for INR analysis, demonstrating its utility across three tasks: (1) a data-driven strategy (SEC-Conf) for hyperparameter selection that outperforms existing heuristics and is robust to model depth, (2) a reliable proxy for signal complexity, and (3) effective alignment of spectral biases across diverse INR architectures.
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2605.12709 [cs.LG] |
| (or arXiv:2605.12709v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2605.12709 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Adam Kania [view email]
[v1]
Tue, 12 May 2026 20:16:48 UTC (4,686 KB)