Active Learning for Conditional Generative Compressed Sensing
Alexander DeLise, Nick Dexter · 2026
Read Active Learning for Conditional Generative Compressed Sensing by Alexander DeLise, Nick Dexter (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-05435.txt.
Introduction
Compressed sensing rebuilds images from very few measurements, guided by a learned generator. Alexander DeLise and Nick Dexter study a conditional version, where a text prompt steers both sampling and recovery. They separate the prompt's two jobs so each can be studied on its own.
For common generator types they prove stable recovery bounds. When the sampling prompt matches the recovery prompt, the method keeps its near-optimal guarantees, while a mismatch adds an explicit penalty. Tests with Stable Diffusion confirm that prompts reshape where samples are taken and change how well images come back.
Catalog record
| CREATOR | Alexander DeLise, Nick Dexter |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 89 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.05435
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Active Learning for Conditional Generative Compressed Sensing Alexander DeLise, Nick Dexter Abstract Generative compressed sensing uses the range of a pretrained generator as a nonlinear model for recovering structured signals from limited measurements. We study a conditional version of this problem for image recovery from subsampled Fourier measurements using prompt-conditioned generative models. Our framework separates two roles of conditioning: the prompt used to design the sampling distribution and the prompt used to define the recovery model. For ReLU and Lipschitz conditional generators, we prove stable recovery bounds showing that prompt-matched Christoffel sampling retains the same Christoffel complexity constant as existing near-optimal generative compressed sensing theory, while prompt mismatch incurs an explicit compatibility penalty. Experiments with Stable Diffusion show that prompts meaningfully reshape Christoffel sampling distributions and influence image recovery. Overall, our results suggest that prompts should be treated as design variables with distinct effects on sensing, approximation, and recovery. 1 Introduction Compressed sensing [ 19 ] is a popular signal processing technique to recover a signal \mathbf{f}^{*}\in\mathbb{C}^{n} from an undersampled observation \mathbf{y}\in\mathbb{C}^{m} , with m\ll n , obtained via some forward process \mathbf{y}=\mathbf{A}\mathbf{f}^{*}+\mathbf{e}, (1) with numerous applications throughout science and engineering [ 12 , 36 , 20 , 53 , 24 ] . Here \mathbf{A}\in\mathbb{C}^{m\times n} is the sampling matrix (with possibly random entries) and \mathbf{e}\in\mathbb{C}^{m} is a vector of additive noise. The inverse problem in Equation 1 is underdetermined. Thus one typically assumes some application-driven structure, usually sparsity of \mathbf{f}^{*} in some domain, to usher a unique solution [ 9 , 10 , 50 , 23 , 41 , 51 , 14 , 3 ] , provided the sampling matrix \mathbf{A} satisfies some condition like the Restricted Isometry Property [ 19 , 10 ] . Generative compressed sensing, on the other hand, replaces the sparsity assumption by instead requiring that \mathbf{f}^{*} lies close to the range of a generative model G:\mathbb{R}^{k}\to\mathbb{R}^{n} , as first introduced in [ 8 ] . In this regime, \mathbf{f}^{*} is recoverable with high probability when the sampling matrix \mathbf{A} with Gaussian entries satisfies the so-called Set-Restricted Eigenvalue Condition. Several subsequent studies investigate upper and lower error bounds for signal recovery in generative compressed sensing [ 28 , 29 , 35 ] as well as recovery guarantees under the presence of outlier data [ 27 ] or with nonlinear sampling operators [ 13 ] . While these results establish a powerful alternative to sparsity-based recovery, they rely on assumptions about the measurement process that are often not met in practice. First, Gaussian measurements can be unrealistic: in many practical applications, sensing is dictated by hardware constraints and corresponds to structured, often subsampled unitary transforms, e.g., the discrete Fourier transform in MRI [ 36 , 37 ] . Second, even after restricting to such physically realizable operators, the choice of sampling pattern remains critical. Common strategies such as Gaussian or uniform subsampling are uninformed, as they fail to exploit the structure of the signal class being measured and may lead to suboptimal performance [ 2 , 33 , 43 ] . Several studies propose sampling strategies for generative compressed sensing that exploit the geometry of the underlying generative model, a process called active learning . For example, [ 1 ] develops the Christoffel Sampling for Machine Learning (CS4ML) framework based on the generalized Christoffel function , while [ 7 , 6 , 42 ] use local coherences to design sampling distributions that place more mass on measurement coordinates that are most informative for distinguishing signals in the model class. Such active learning approaches relate closely to the well-studied leverage score sampling , with various applications in active learning, standard regression, and imaging [ 15 , 18 , 40 , 21 , 5 ] , as well optimal sampling strategies for compressed sensing in general [ 3 , 33 , 43 , 11 , 34 , 44 ] . Despite these advances, existing work in generative compressed sensing has largely relied on unconditional generative priors, thereby omitting auxiliary information that may be available at acquisition or inference time. In many realistic settings, such information can sharply restrict the set of plausible signals. For example, in medical imaging, modality, anatomy, or patient metadata may improve reconstruction [ 16 ] ; similarly, in natural image reconstruction, subject, scene category, acquisition protocol, or other metadata may narrow the relevant signal class and improve reconstruction quality [ 17 , 30 ] . Prompted or otherwise conditional generators provide a natural mechanism for incorporating this information by steering the recovery class toward signals compatible with the available context, rather than requiring a single unconditional model to represent all possible signal types. Conditioning can also improve sensing itself. If the relevant conditional signal class is narrower, then one may design sampling distributions supported on more informative measurement coordinates, leading to lower sample complexity for stable recovery. However, conditioning introduces interactions absent from the unconditional setting: the prompt used for sampling need not match the prompt used for recovery, and neither need match the class containing the true signal, when \mathbf{f}^{*} lies in the range of the generator. Thus, overly strong or misaligned conditioning may bias the reconstruction toward the prior and away from the observed measurements. This motivates conditional generative compressed sensing as an extension of the standard framework in which both sampling design and recovery guarantees depend explicitly on the interaction between conditional signal classes. Understanding how these interactions affect stable recovery and reconstruction is the central question of this paper. 1.1 Contributions We adapt an active learning perspective via Christoffel sampling from [ 1 ] together with generative compressed sensing arguments from [ 8 , 7 , 6 ] to study prompt-conditioned signal recovery under subsampled Fourier measurements. The resulting theory characterizes how interactions between the sampling, recovery, and true-signal prompts determine the sample complexity required for stable recovery and the resulting reconstruction error. Our contributions may be summarized as follows: • We formulate a prompt-conditioned Christoffel sampling framework for generative compressed sensing under subsampled Fourier measurements with three conditioning prompts: the true-signal prompt c_{*} , the recovery prompt c_{r} , and the sampling prompt c_{s} . • We identify a single prompt compatibility factor \Lambda(c_{1},c_{2},c_{3}) and show that evaluating it at (c_{r},c_{r},c_{s}) controls sample complexity for stable reconstruction, while evaluating it at (c_{*},c_{r},c_{s}) controls the residual error in recovery. • We derive explicit sample complexity results under ReLU and Lipschitz assumptions on the conditional generator and combine these with agnostic signal recovery error bounds. • We empirically test the theoretical framework through a suite of experiments demonstrating that prompt conditioning systematically influences both the induced sampling distributions and reconstruction performance. 2 Active Learning Problem Setup We follow the general Christoffel sampling for machine learning (CS4ML) framework established in [ 1 ] for our active learning regime, however we specialize our results to subsampled Fourier measurements. For a full description of the general CS4ML framework, we refer the reader to Appendix A . 2.1 Setup and Prompt-Conditioned Model Classes Let \mathbb{X}=\mathbb{R}^{n} , deemed the object space , be equipped with Euclidean norm \left\|\cdot\right\|_{2} . Let \mathbf{F}\in\mathbb{C}^{n\times n} denote the unitary discrete Fourier transform, let D=\{1,\dots,n\} , and let \mathbf{P}_{i} be the row selector matrix for the i th Fourier coefficient. For