Deep Compressed Sensing

Abstract

Compressed sensing (CS) provides an elegant framework for recovering sparse signals from compressed measurements. For example, CS can exploit the structure of natural images and recover an image from only a few random measurements. CS is flexible and data efficient, but its application has been restricted by the strong assumption of sparsity and costly reconstruction process. A recent approach that combines CS with neural network generators has removed the constraint of sparsity, but reconstruction remains slow. Here we propose a novel framework that significantly improves both the performance and speed of signal recovery by jointly training a generator and the optimisation process for reconstruction via meta-learning. We explore training the measurements with different objectives, and derive a family of models based on minimising measurement errors. We show that Generative Adversarial Nets (GANs) can be viewed as a special case in this family of models. Borrowing insights from the CS perspective, we develop a novel way of improving GANs using gradient information from the discriminator.

ICML 2019

Cited by in corpus (26)

• Review: Deep Learning in Electron Microscopy
• Algorithmic Guarantees for Inverse Imaging with Untrained Network Priors
• LOGAN: Latent Optimisation for Generative Adversarial Networks
• Joint Design of Measurement Matrix and Sparse Support Recovery Method via Deep Auto-encoder
• Optimal Transport driven CycleGAN for Unsupervised Learning in Inverse Problems
• Self-Supervision with Superpixels: Training Few-shot Medical Image Segmentation without Annotation
• Bayesian Inference with Generative Adversarial Network Priors
• Model-Aware Deep Architectures for One-Bit Compressive Variational Autoencoding
• Globally Injective ReLU Networks
• Latent-Optimized Adversarial Neural Transfer for Sarcasm Detection
• Spark Deficient Gabor Frame Provides a Novel Analysis Operator for Compressed Sensing
• Enhanced Few-shot Learning for Intrusion Detection in Railway Video Surveillance
• FOAL: Fast Online Adaptive Learning for Cardiac Motion Estimation
• Generalized Energy Based Models
• CSMCNet: Scalable Video Compressive Sensing Reconstruction with Interpretable Motion Estimation
• Scalable Deep Compressive Sensing
• Gradient Importance Learning for Incomplete Observations
• OpenICS: Open Image Compressive Sensing Toolbox and Benchmark
• Learning Generative Prior with Latent Space Sparsity Constraints
• Model-Aware Regularization For Learning Approaches To Inverse Problems
• Just Least Squares: Binary Compressive Sampling with Low Generative Intrinsic Dimension
• Generative Model Adversarial Training for Deep Compressed Sensing
• Joint User Activity and Data Detection in Grant-Free NOMA using Generative Neural Networks
• Compressed Sensing via Measurement-Conditional Generative Models
• Non-Convex Compressed Sensing with Training Data
• Meta-learning for RIS-assisted NOMA Networks