Newest papers

Training-free Acceleration for Blind Compressed Image Restoration via Activation-Calibrated Singular Value Decomposition

姚苏杭, Xin Li, Xijun Wang, Yufei Wang, Yeying Jin, Zhibo Chen · 2026 · IEEE Journal on Emerging and Selected Topics in Circuits and Systems

DOI: 10.1109/jetcas.2026.3745896 · cited by 0 works (OpenAlex) · open-access PDF

Spanish

Los modelos activos basados en transformer 21 han reportado un fuerte desempeño en la restauración de imágenes comprimidas a ciegas (CIR) bajo configuraciones de compresión desconocidas, pero sus operadores de proyección densamente empacados hacen la implementación costosa. Los modelos de transformer 21 han estudiado la aceleración sin entrenamiento para tales modelos a través de un análisis empírico, de nivel de operador. En concreto, perfilamos las vías de reconstrucción ben-condicionantes donde se concentra la computación, caracterizamos las distribuciones de activación en dirección de canal bajo las entradas de compresión originales, y examinamos la decadencia de valor singular de los pesos de proyección representativos. Estas mediciones identifican las proyecciones recurrentes en la atención, el alimento adelante, la convolución y el prompt-condicionamiento

Machine translation (Meta M2M100); the English original follows.

Abstract

Transformer-based models achieve strong performance on blind compressed image restoration (CIR) under unknown compression settings, but their densely stacked projection operators make deployment costly. We study training-free acceleration for such models through an empirical, operator-level analysis. Specifically, we profile where computation is concentrated, characterize channel-wise activation distributions under compressed inputs, and examine the singular-value decay of representative projection weights. These measurements identify the recurring projections in the attention, feed-forward, convolutional, and prompt-conditioning pathways as effective low-rank replacement targets that preserve the original inference graph. They further establish truncated SVD (TSVD) as a principled training-free baseline for blind CIR and reveal why a uniform low-rank objective is mismatched to the channel-imbalanced feature distributions of compressed images. Accordingly, we formulate Activation-Calibrated Singular Value Decomposition (ASVD) for this restoration setting, deriving low-rank factors entirely offline from pretrained weights and lightweight activation statistics gathered from only a few compressed calibration images. By weighting the decomposition with channel-wise activation magnitudes, our ASVD assigns larger penalties to errors on more active channels, requiring neither retraining nor online calibration and leaving the restoration pipeline intact. When instantiated on PromptCIR at rank 64, ASVD consistently improves PSNR-B over TSVD across four standard benchmarks under identical parameter and MAC budgets. At this operating point, the factorized model cuts parameters by 59% (34.75M to 14.34M) and computation by 16.1% (258.79 to 217.16 GMACs for a 256 x 256 input), enabling acceleration without retraining.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Singular value decomposition, Projection (relational algebra), Computation, Acceleration, Pipeline (software), Compressed sensing

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange