中文

The Singularity Space: A Generative Diffusion Framework for Signal Representation

机器学习 2026-07-12 v1 人工智能 信号处理 数值分析

摘要

Generative models often represent signals as dense grids of amplitudes, blurring sharp transients that are crucial for the correctness of physical signals. We introduce Singularity Space, a generative framework that represents signals through complex-plane singularities, rooted in the classical pole-residue representation of meromorphic functions. We learn a latent space of physically constrained, per-signal singularity configurations to solve an inverse problem from degraded or partial observations. The framework has three key properties: interpretability, in which each generated singularity configuration corresponds to a set of physical parameters; structural stability, which mitigates Gibbs artifacts at discontinuities; and resolution-free output reconstruction on arbitrary grids without retraining or interpolation. Our framework employs a transformer-based diffusion model that directly predicts samples at complex-plane singularity coordinates, subject to geometric constraints during sampling. As a controlled test case for sharp-feature recovery, we evaluate our framework on 1D Burgers shocks, where each shock is represented by 32 predicted singularities (an 8×8\times reduction versus a 1024-point grid signal). Our framework preserves signal structure (TV ratio1\text{TV ratio} \approx 1) under unseen test-time observation noise, achieves a 4.2×4.2\times lower reconstruction error in zero-shot sub-resolution generalization than a grid-based baseline, and recovers physical parameters to 10410^{-4} absolute error in-distribution. These results suggest that singularity-based representations may provide a practical foundation for other transient-dominated signals such as speech and biomedical signals, with potential extension to higher-dimensional domains.

引用

@article{arxiv.2607.10930,
  title  = {The Singularity Space: A Generative Diffusion Framework for Signal Representation},
  author = {Eli Bar-Yosef and Amir Averbuch and Eli Turkel},
  journal= {arXiv preprint arXiv:2607.10930},
  year   = {2026}
}