Denoise-then-Retrieve Paradigm

Updated 8 July 2026

Denoise-then-retrieve paradigm is a workflow that first purifies input data by removing nuisance signals and then applies retrieval operations to recover accurate outputs.
It encompasses strategies like conditional GAN-based denoising, text-conditioned filtering, and implicit denoiser priors that tailor noise suppression to specific downstream tasks.
Empirical studies in medical imaging, video retrieval, and phase reconstruction show significant performance boosts, validating its practical efficacy in diverse applications.

Searching arXiv for papers on the denoise-then-retrieve paradigm and adjacent formulations. The denoise-then-retrieve paradigm is a family of workflows in which nuisance content is suppressed before, or as part of, a downstream retrieval operator. Across the supplied literature, “denoising” is not restricted to pixel smoothing: it can mean projecting an observation onto a learned noise manifold, filtering text-irrelevant video clips, compressing retrieved evidence into claim-relevant context, or using a denoiser-derived implicit prior during inverse reconstruction. The unifying principle is that retrieval quality degrades when the retriever is forced to operate on representations dominated by irrelevant, corrupted, or semantically mismatched information, and that performance can improve when the input to retrieval is first purified or selectively constrained (Liu et al., 15 Aug 2025).

1. Conceptual definition and scope

Within the supplied sources, the paradigm appears in both strict and hybrid forms. In its strict form, denoising is a distinct upstream stage that transforms the observation before a conventional retrieval or reconstruction step. The SPECT pipeline is an explicit example: a conditional GAN denoises sinograms, and only then are the cleaned sinograms reconstructed using Ordered Subset Expectation Maximization (OSEM) (Chrysostomou, 2021). The video moment retrieval formulation is similarly explicit: text-irrelevant clips are filtered first, and conditional retrieval is then performed on purified multimodal representations (Liu et al., 15 Aug 2025).

A broader interpretation treats denoising as representation purification rather than signal restoration. In zero-shot composed image retrieval, Denoise-I2W first maps a reference image into a denoising pseudo-word token that excludes intention-irrelevant visual information, and retrieval is subsequently performed with a more faithful composed query representation (Tang et al., 2024). In phase retrieval, the denoiser may operate as an implicit learned prior or as a regularizer within an iterative solver, so denoising and retrieval become interleaved rather than strictly sequential (Agrawal et al., 2022); (Wang et al., 2020).

The supplied papers also show that the paradigm has neighboring but distinct forms. Decide Then Retrieve is described as a training-free framework that decides before retrieving, and more precisely as a hybrid that performs denoising both before retrieval and during retrieval selection rather than a pure denoise-then-retrieve pipeline (Chen et al., 7 Jan 2026). Retrieve-Refine-Calibrate is even more clearly retrieve-then-denoise-then-verify: evidence is retrieved first, then refined to remove irrelevant information, and only after that is verification calibrated (Sun et al., 23 Jan 2026). This suggests that the paradigm is best understood as a general design principle about controlling nuisance information, with different papers choosing different insertion points for the denoising operation.

2. Measurement-space denoising before reconstruction

A canonical formulation appears in imaging pipelines where retrieval means reconstructing a latent signal from noisy measurements. In SPECT imaging, the proposed method generates noisy sinograms from software phantoms by forward projection and Poisson corruption, trains a conditional GAN to map noisy sinograms to denoised sinograms using paired noisy/clean data, and reconstructs images with OSEM from either the original noisy sinograms or the GAN-denoised sinograms (Chrysostomou, 2021). The retrieval stage is therefore not document ranking but tomographic recovery of the final image.

The architecture in that setting is a conditional GAN with an encoder-decoder generator and a discriminator trained to distinguish generated denoised sinograms from true clean sinograms. The generator is conditioned on the noisy sinogram, uses skip connections, and outputs a denoised sinogram; the discriminator has three convolutional layers and a final fully connected sigmoid output (Chrysostomou, 2021). The paper reports denoising results on the Shepp-Logan-based evaluation for low, medium, and high noise, and compares standard OSEM reconstructions against OSEM applied to GAN-denoised sinograms.

Setting	Metric summary	Reported values
Sinogram denoising	Low noise	MAPE 4.80%, MSE 0.0009, SSIM 0.975, PSNR 30.48
Sinogram denoising	Medium noise	MAPE 5.83%, MSE 0.0010, SSIM 0.971, PSNR 29.91
Sinogram denoising	High noise	MAPE 8.39%, MSE 0.0054, SSIM 0.930, PSNR 22.65

For the reconstruction stage, the paper reports that standard OSEM at low noise yields MSE 0.0082, SSIM 0.82, and PSNR 20.89, while the proposed denoise-then-retrieve pipeline yields MSE 0.0040, SSIM 0.89, and PSNR 23.94. At medium noise, the corresponding values are 0.0087/0.80/20.58 for standard OSEM and 0.0045/0.89/23.48 for the proposed method; at high noise, 0.0099/0.75/20.05 versus 0.0061/0.84/22.17 (Chrysostomou, 2021). The authors state that the proposed method “significantly denoise[s] the sinograms” and “significantly improves the reconstructions.”

This formulation captures the most literal interpretation of the paradigm: improve the observation first, then apply the unchanged retrieval operator. A plausible implication is that such pipelines are especially attractive when a well-established downstream inverse method already exists and the principal bottleneck is observation quality rather than reconstruction architecture.

3. Representation denoising in multimodal retrieval

In text-driven Video Moment Retrieval, the denoise-then-retrieve formulation targets semantic noise rather than sensor noise. The paper argues that standard methods encode all video clips, including irrelevant ones, disrupting multimodal alignment and hindering optimization. DRNet operationalizes the alternative by first filtering text-irrelevant clips from videos and then retrieving the target moment using purified multimodal representations (Liu et al., 15 Aug 2025).

The first core module is Text-Conditioned Denoising (TCD). Video clips are encoded by concatenating CLIP and SlowFast features, text by the CLIP text encoder, and cross-attention injects query information into each video clip. A Context Interaction Operator built from bidirectional Mamba blocks then performs multimodal context integration. From the resulting contextualized features, dynamic kernels derived from pooled text features compute clip-wise text-clip alignment scores, which are thresholded with $\mu$ to form a binary noise mask. Applying this mask yields purified visual features $\hat{\boldsymbol f_v}$ (Liu et al., 15 Aug 2025).

The second module, Text-Reconstruction Feedback (TRF), provides auxiliary supervision. TRF reconstructs a sentence-level query embedding from the purified visual representation and aligns it with the pooled text embedding via a cosine-distance semantic consistency loss with $\lambda_t=2$ . The paper explicitly frames this as preventing overly aggressive masking: if the purified video cannot reconstruct the text, the mask is probably too harsh or inaccurate (Liu et al., 15 Aug 2025). Retrieval is then performed on the purified multimodal representation rather than on the raw clip sequence.

The reported results on QVHighlights test are 66.73 [email protected], 50.52 [email protected], 64.17 mAP@0.5, 45.79 [email protected], and 43.73 mAP@Avg. On Charades-STA, DRNet reaches 60.86 [email protected] and 39.78 [email protected] with SlowFast+CLIP features, and 69.75 mIoU with VGG features; with VGG it also reports 59.03 [email protected] and 36.26 [email protected] (Liu et al., 15 Aug 2025). The ablations attribute the largest drop to removing TCD, while removing TRF also hurts performance, and replacing Mamba-based CIOs with standard Transformers leads to a noticeable drop.

An analogous but distinct form appears in zero-shot composed image retrieval. Denoise-I2W addresses the mismatch that ordinary image-caption pre-training maps a reference image into a pseudo-word token that often carries intention-irrelevant content. Its pseudo triplet construction module automatically creates triples of pseudo-reference image, pseudo-manipulation text, and target image; its pseudo-composed mapping module then trains the mapping network to produce a denoising pseudo-word token that, when composed with the manipulation text, aligns with the target image (Tang et al., 2024). The method is described as model-agnostic and annotation-free.

Task	Baseline-to-Denoise-I2W change	Reported values
CIRR with Context-I2W	R@1/R@5/R@10/R@50	25.6/55.1/68.5/89.8 to 26.9/57.2/69.8/90.6
ImageNet domain conversion	Avg. R@10/R@50	12.9/27.6 to 14.3/30.6
COCO object composition	R@1/R@5/R@10	13.5/28.5/38.1 to 16.1/33.0/43.5

The paper further states that integrating Denoise-I2W with existing best models obtains consistent and significant performance boosts ranging from 1.45% to 4.17% without increasing inference costs (Tang et al., 2024). In both DRNet and Denoise-I2W, denoising is a query-conditioned suppression of semantically irrelevant content before retrieval, not low-level restoration.

4. Denoiser priors in phase retrieval

Phase retrieval provides a more mathematical instance of the paradigm, where “retrieve” refers to recovering a signal from Fourier magnitude measurements. The supplied papers present two related formulations. Alternating Phase Langevin Sampling (APLS) treats the denoiser as an implicit prior and alternates between denoiser-guided posterior sampling and phase estimation (Agrawal et al., 2022). “When deep denoising meets iterative phase retrieval” embeds a deep denoiser inside ADMM/HIO-style splitting via Regularization-by-Denoising (RED), creating hybrid denoise-then-retrieve and denoise-while-retrieve algorithms (Wang et al., 2020).

APLS starts from the phase retrieval model

$b = |Ax^*| + w,$

and for the Fourier case uses a noisy Fourier magnitude model with $4\times$ oversampled Fourier measurements. The outer loop updates the phase by $p_t \gets \mathrm{Ph}(Ax_{t-1})$ , forms phase-augmented linearized measurements $x^c = p_t \odot b$ , and calls a constrained Langevin posterior sampler with update

$x \gets x + h d_x + \gamma z,$

where

$d_x \gets (I - A^T A)D(x) + A^T(x^c - Ax).$

This is precisely described as a Denoise-then-Retrieve method: first denoise or sample using the implicit denoiser prior, then retrieve the phase-consistent estimate, and continue alternating (Agrawal et al., 2022).

The reported results emphasize robustness to distribution shift. On natural images at $\alpha=2$ , HIO scores 18.72, prDeep 30.02, and APLS 29.34; at $\hat{\boldsymbol f_v}$ 0, 19.04, 28.62, and 28.69; at $\hat{\boldsymbol f_v}$ 1, 18.56, 28.05, and 27.67. On unnatural images at $\hat{\boldsymbol f_v}$ 2, HIO scores 19.75, prDeep 21.9, and APLS 22.32; at $\hat{\boldsymbol f_v}$ 3, 21.24, 24.08, and 25.5; at $\hat{\boldsymbol f_v}$ 4, 18.57, 22.29, and 22.44. On the Shenzhen chest X-ray dataset at $\hat{\boldsymbol f_v}$ 5, prDeep achieves average PSNR 33.68, while APLS reaches 35.04 (Agrawal et al., 2022).

The RED-based formulation begins from

$\hat{\boldsymbol f_v}$ 6

and uses the RED regularizer

$\hat{\boldsymbol f_v}$ 7

The paper proposes RED-ITA-F and RED-ITA-S, both ADMM-based hybrids that retain the physics of iterative phase retrieval while injecting a learned prior through DnCNN denoisers (Wang et al., 2020). RED-ITA-F relaxes the Fourier measurement term; RED-ITA-S relaxes the oversampling or signal-consistency constraint. The paper explicitly states that with random initialization, Deep-ITA-F/S achieve the best PSNR and SSIM by large margins, and under random init and $\hat{\boldsymbol f_v}$ 8, Deep-ITA methods achieve around 61 dB PSNR overall versus 28 dB for DnCNN-ADMM and 24–34 dB for HIO/OSS/prDeep (Wang et al., 2020).

A related but more general denoising formulation appears in “Reconstructing the Noise Manifold for Image Denoising,” which argues that it is easier and more reliable to model the residual image than the full clean image space, and therefore better to constrain a denoiser to remove only information spanning the manifold of the residual image (Marras et al., 2020). The paper does not perform retrieval itself, but the supplied details describe it as a denoise-then-retrieve-compatible denoiser because it is explicitly designed to preserve image content and structure while removing realistic sensor noise. Reported results include, on DnD, 49.90 dB/0.9861 RAW and 41.50 dB/0.9759 sRGB for the blind ResNet backbone, and 50.91 dB/0.9873 RAW and 42.11 dB/0.9775 sRGB for the non-blind ResNet backbone (Marras et al., 2020).

Taken together, these papers show that in inverse problems the paradigm can mean either a sequential denoise-then-solve recipe or an alternating algorithm in which denoising acts as a learned prior inside the retrieval loop.

Not all systems in the supplied literature are pure denoise-then-retrieve methods, but several are closely related because they interpret denoising as filtering retrieval necessity or evidence quality. Decide Then Retrieve (DTR) changes the standard RAG order: the model first attempts a parametric answer, computes a generation uncertainty score

$\hat{\boldsymbol f_v}$ 9

and triggers retrieval only when uncertainty exceeds a threshold such as $\lambda_t=2$ 0 (Chen et al., 7 Jan 2026). The paper states that no-retrieval accuracy drops monotonically as uncertainty increases, validating uncertainty as a proxy for parametric correctness.

DTR then uses dual-path retrieval with adaptive information selection. One path retrieves with the original query embedding, the other with a pseudo-context generated by the prompt “Write a passage to answer this question,” and the candidate documents are scored jointly by

$\lambda_t=2$ 1

The paper further gives a geometry-motivated rule

$\lambda_t=2$ 2

The supplied description explicitly characterizes this as a hybrid rather than a pure denoise-then-retrieve method: DTR performs denoising or filtering at the decision stage by avoiding retrieval for low-uncertainty queries and at the retrieval stage by suppressing passages not jointly supported by both views (Chen et al., 7 Jan 2026).

Retrieve-Refine-Calibrate (RRC) positions the denoising step after retrieval. Instead of decomposing a claim into sub-claims, it extracts up to three entities with LUKE, retrieves evidence with BM25, refines the retrieved evidence into a concise claim-relevant paragraph $\lambda_t=2$ 3, and calibrates low-confidence predictions using thresholds $\lambda_t=2$ 4 for HOVER and $\lambda_t=2$ 5 for FEVEROUS-S (Sun et al., 23 Jan 2026). The paper’s central claim is that decomposition can inject incorrect intermediate entities or relations, so the bottleneck is noisy intermediate representations rather than weak reasoning ability.

Framework	Retrieval order	Denoising operation
DTR	Decide before retrieval	Uncertainty-guided triggering and adaptive evidence selection
RRC	Retrieve before denoising	Claim-specific evidence refinement and selective calibration

The main results reported for RRC are Macro-F1 scores of 69.33 on HOVER 2-hop, 63.98 on HOVER 3-hop, 60.95 on HOVER 4-hop, and 72.55 on FEVEROUS-S (Sun et al., 23 Jan 2026). Ablations show drops when removing the evidence refinement module, supporting the claim that retrieval noise is a major problem in complex claims. These systems broaden the paradigm: denoising can target the decision to retrieve, the retrieved evidence itself, or both.

6. Common principles, misconceptions, and boundaries

Across the supplied papers, several common principles recur. First, denoising is usually task-conditioned. DRNet’s mask is conditioned on text and suppresses text-irrelevant clips (Liu et al., 15 Aug 2025). Denoise-I2W keeps only image information useful for the manipulation intent (Tang et al., 2024). RRC refines evidence with respect to the original claim rather than using generic compression (Sun et al., 23 Jan 2026). This suggests that the paradigm is not merely about reducing variance; it is about removing information that is irrelevant for a specific downstream retrieval objective.

Second, the denoising stage is often constrained so that useful content is not discarded. DRNet uses TRF to ensure purified video still reconstructs sentence semantics (Liu et al., 15 Aug 2025). The noise-manifold denoiser argues that the generator should remove only information spanning the manifold of the residual image, which is presented as a way to reduce oversmoothing and preserve detail (Marras et al., 2020). In phase retrieval, the denoiser is balanced against measurement consistency rather than used as a free-standing image translator (Agrawal et al., 2022); (Wang et al., 2020).

Third, a common misconception is that denoise-then-retrieve always denotes a simple prefilter followed by a fixed retriever. The supplied literature contradicts that narrow reading. In APLS, denoising and retrieval alternate inside a single loop (Agrawal et al., 2022). In RED-ITA, the denoiser is part of the optimization objective itself (Wang et al., 2020). In DTR and RRC, denoising is implemented as decision filtering or evidence refinement rather than observation restoration (Chen et al., 7 Jan 2026); (Sun et al., 23 Jan 2026).

A further boundary condition concerns evidentiary support. One supplied record, “Reason-Then-Retrieve for CoVR-R with Structured Edit Prompts and Dense-Sparse Fusion,” includes an abstract describing a reason-then-retrieve pipeline and reported retrieval metrics, but the accompanying details state that the provided content is only a LaTeX rebuttal or template document and therefore does not contain the method body or retrieval formulas (Liu et al., 1 Jun 2026). From the supplied text alone, its relation to a denoise-then-retrieve interpretation cannot be specified beyond noting that it belongs to an adjacent family in which reasoning precedes retrieval.

Overall, the denoise-then-retrieve paradigm denotes a methodological commitment: retrieval should operate on purified inputs, purified representations, or purified evidence rather than on raw, noise-dominated signals. The supplied literature shows this commitment instantiated in medical imaging, phase retrieval, video moment retrieval, zero-shot composed image retrieval, adaptive retrieval-augmented generation, and fact-checking, with the exact locus of denoising varying by task and model class (Chrysostomou, 2021).