OmniSpectra：通用光谱表征学习的统一基础模型
OmniSpectra: A Unified Foundation Model for Universal Spectra Representation Learning
AI 导读
研究者提出 OmniSpectra，一个面向光谱表征学习的波长原生基础模型，采用连续坐标掩码自编码器取代整数像素索引，并以分辨率门控的 Fourier 解码器抗混叠。模型在八个巡天数据上预训练，与七个基线对比后，仅以 12.2M 参数达到 SOTA，超过规模大得多的模型，表明编码物理结构可替代参数规模。
正文 · 原文
Abstract:Standard sequence models encode position by token rank rather than physical coordinate, typically forcing practitioners to destructively resample continuous physical signals such as spectra observed across heterogeneous instruments. We identify this coordinate mismatch as a structural limitation in cross-instrument representation learning, and propose OmniSpectra, a wavelength-native foundation model for spectral representation learning: a continuous-coordinate masked autoencoder that discards the integer pixel indexing of sequence modeling. Mapping physical coordinates to a logarithmic basis turns the multiplicative scaling induced by cosmological redshift into a uniform additive shift. In continuous rotary position embeddings, this shared shift cancels from relative token distances rather than producing grid-dependent positional distortion. We further counter the spectral bias inherent to implicit neural decoders with a physics-grounded, resolution-gated Fourier decoder that provides anti-aliasing. Pretrained on eight surveys and benchmarked against seven baselines, our model reaches state-of-the-art performance with only 12.2M parameters while outperforming substantially larger models, indicating that encoding physical structure can substitute for parameter scale.
| Subjects: | Instrumentation and Methods for Astrophysics (astro-ph.IM); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2601.15351 [astro-ph.IM] |
| (or arXiv:2601.15351v2 [astro-ph.IM] for this version) | |
| https://doi.org/10.48550/arXiv.2601.15351 arXiv-issued DOI via DataCite |
Submission history
From: Md Khairul Islam [view email]
[v1]
Wed, 21 Jan 2026 04:39:32 UTC (1,099 KB)
[v2]
Fri, 23 Oct 2026 00:55:35 UTC (2,668 KB)
来源：arXiv cs.AI · arxiv.org