Dual-Ascent-Inspired Transformer for Compressed Sensing

TL;DR

A new transformer-based model for image compressed sensing adapts to different compression ratios with minimal training and high performance.

Contribution

The Dual-Ascent-Inspired Transformer (DAT) maintains stable performance across varying compression ratios with low training costs.

Findings

DAT outperforms existing methods in early-stage training across multiple compression ratios.

DAT achieves comparable PSNR to ISTA-Net+ within one epoch, while others need more training time.

DAT shows robustness to initial learning rate variations during training.

Abstract

Deep learning has revolutionized image compressed sensing (CS) by enabling lightweight models that achieve high-quality reconstruction with low latency. However, most deep neural network-based CS models are pre-trained for specific compression ratios (CS ratios), limiting their flexibility compared to traditional iterative algorithms. To address this limitation, we propose the Dual-Ascent-Inspired Transformer (DAT), a novel architecture that maintains stable performance across different compression ratios with minimal training costs. DAT’s design incorporates the mathematical properties of the dual ascent method (DAM), leading to accelerated training convergence. The architecture features an innovative asymmetric primal–dual space at each iteration layer, enabling dimension-specific operations that balance reconstruction quality with computational efficiency. We also optimize the Cross Attention module through parameter sharing, effectively reducing its training complexity. Experimental results demonstrate DAT’s superior performance in two key aspects: First, during early-stage training (within 10 epochs), DAT consistently outperforms existing methods across multiple CS ratios (10%, 30%, and 50%). Notably, DAT achieves comparable PSNR to the ISTA-Net+ baseline within just one epoch, while competing methods require significantly more training time. Second, DAT exhibits enhanced robustness to variations in initial learning rates, as evidenced by loss function analysis during training.

Figures