PyTorchKR Papers

최신 논문

총 547편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2206.13618공식 코드 · Papers with Code

Fast Low Rank column-wise Compressive Sensing for Accelerated Dynamic MRI

This work develops a fast, memory-efficient, and general algorithm for accelerated/undersampled dynamic MRI by assuming an approximate LR model on the matrix formed by the vectorized images of the sequence. By general, we mean that our algorithm can be used for multiple accelerated dynamic MRI applications and multiple sampling rates (acceleration rates) and patterns with a single choice of parameters (no parameter tuning). We show that our proposed algorithms, alternating Gradient Descent (GD) and minimization for MRI (altGDmin-MRI and altGDmin-MRI2), outperform many existing approaches while also being faster than all of them, on average. This claim is based on comparisons on 8 different retrospectively undersampled single- or multi-coil dynamic MRI applications, undersampled using either 1D Cartesian or 2D pseudo-radial undersampling at multiple sampling rates. All comparisons used the same set of algorithm parameters. Our second contribution is a mini-batch and a fully online extension that can process new measurements and return reconstructions either as soon as measurements of a new image frame arrive, or after a short delay.

arXiv:2206.11311

On Grid Compressive Sampling for Spherical Field Measurements in Acoustics

We derive a compressive sampling method for acoustic field reconstruction using field measurements on a predefined spherical grid that has theoretically guaranteed relations between signal sparsity, measurement number, and reconstruction accuracy. This method can be used to reconstruct band-limited spherical harmonic or Wigner $D$-function series (spherical harmonic series are a special case) with sparse coefficients. Contrasting typical compressive sampling methods for Wigner $D$-function series that use arbitrary random measurements, the new method samples randomly on an equiangular grid, a practical and commonly used sampling pattern. Using its periodic extension, we transform the reconstruction of a Wigner $D$-function series into a multi-dimensional Fourier domain reconstruction problem. We establish that this transformation has a bounded effect on sparsity level and provide numerical studies of this effect. We also compare the reconstruction performance of the new approach to classical Nyquist sampling and existing compressive sampling methods. In our tests, the new compressive sampling approach performs comparably to other guaranteed compressive sampling approaches and needs a fraction of the measurements dictated by the Nyquist sampling theorem. Moreover, using one-third of the measurements or less, the new compressive sampling method can provide over 20 dB better denoising capability than oversampling with classical Fourier theory.

arXiv:2206.05647

A Fast Alternating Minimization Algorithm for Coded Aperture Snapshot Spectral Imaging Based on Sparsity and Deep Image Priors

Coded aperture snapshot spectral imaging (CASSI) is a technique used to reconstruct three-dimensional hyperspectral images (HSIs) from one or several two-dimensional projection measurements. However, fewer projection measurements or more spectral channels leads to a severly ill-posed problem, in which case regularization methods have to be applied. In order to significantly improve the accuracy of reconstruction, this paper proposes a fast alternating minimization algorithm based on the sparsity and deep image priors (Fama-SDIP) of natural images. By integrating deep image prior (DIP) into the principle of compressive sensing (CS) reconstruction, the proposed algorithm can achieve state-of-the-art results without any training dataset. Extensive experiments show that Fama-SDIP method significantly outperforms prevailing leading methods on simulation and real HSI datasets.

arXiv:2206.04797

Memory-efficient model-based deep learning with convergence and robustness guarantees

Computational imaging has been revolutionized by compressed sensing algorithms, which offer guaranteed uniqueness, convergence, and stability properties. Model-based deep learning methods that combine imaging physics with learned regularization priors have emerged as more powerful alternatives for image recovery. The main focus of this paper is to introduce a memory efficient model-based algorithm with similar theoretical guarantees as CS methods. The proposed iterative algorithm alternates between a gradient descent involving the score function and a conjugate gradient algorithm to encourage data consistency. The score function is modeled as a monotone convolutional neural network. Our analysis shows that the monotone constraint is necessary and sufficient to enforce the uniqueness of the fixed point in arbitrary inverse problems. In addition, it also guarantees the convergence to a fixed point, which is robust to input perturbations. We introduce two implementations of the proposed MOL framework, which differ in the way the monotone property is imposed. The first approach enforces a strict monotone constraint, while the second one relies on an approximation. The guarantees are not valid for the second approach in the strict sense. However, our empirical studies show that the convergence and robustness of both approaches are comparable, while the less constrained approximate implementation offers better performance. The proposed deep equilibrium formulation is significantly more memory efficient than unrolled methods, which allows us to apply it to 3D or 2D+time problems that current unrolled algorithms cannot handle.

arXiv:2206.03572

Compressive Sensing with Wigner $D$-functions on Subsets of the Sphere

In this paper, we prove a compressive sensing guarantee for restricted measurement domains on the rotation group, $\mathrm{SO}(3)$. We do so by first defining Slepian functions on a measurement sub-domain $R$ of the rotation group $\mathrm{SO}(3)$. Then, we transform the inverse problem from the measurement basis, the bounded orthonormal system of band-limited Wigner $D$-functions on $\mathrm{SO}(3)$, to the Slepian functions in a way that limits increases to signal sparsity. Contrasting methods using Wigner $D$-functions that require measurements on all of $\mathrm{SO}(3)$, we show that the orthogonality structure of the Slepian functions only requires measurements on the sub-domain $R$, which is select-able. Due to the particulars of this approach and the inherent presence of Slepian functions with low concentrations on $R$, our approach gives the highest accuracy when the signal under study is well concentrated on $R$. We provide numerical examples of our method in comparison with other classical and compressive sensing approaches. In terms of reconstruction quality, we find that our method outperforms the other compressive sensing approaches we test and is at least as good as classical approaches but with a significant reduction in the number of measurements.

arXiv:2206.02899

Machine Learning Prediction for Phase-less Millimeter-Wave Beam Tracking

Future wireless networks may operate at millimeter-wave (mmW) and sub-terahertz (sub-THz) frequencies to enable high data rate requirements. While large antenna arrays are critical for reliable communications at mmW and sub-THz bands, these antenna arrays would also mandate efficient and scalable initial beam alignment and link maintenance algorithms for mobile devices. Low-power phased-array architectures and phase-less power measurements due to high frequency oscillator phase noise pose additional challenges for practical beam tracking algorithms. Traditional beam tracking protocols require exhaustive sweeps of all possible beam directions and scale poorly with high mobility and large arrays. Compressive sensing and machine learning designs have been proposed to improve measurement scaling with array size but commonly degrade under hardware impairments or require raw samples respectively. In this work, we introduce a novel long short-term memory (LSTM) network assisted beam tracking and prediction algorithm utilizing only phase-less measurements from fixed compressive codebooks. We demonstrate comparable beam alignment accuracy to state-of-the-art phase-less beam alignment algorithms, while reducing the average number of required measurements over time.

arXiv:2206.02510

Single pixel imaging at high pixel resolutions

The usually reported pixel resolution of single pixel imaging (SPI) varies between $32 \times 32$ and $256 \times 256$ pixels falling far below imaging standards with classical methods. Low resolution results from the trade-off between the acceptable compression ratio, the limited DMD modulation frequency, and reasonable reconstruction time, and has not improved significantly during the decade of intensive research on SPI. In this paper we show that image measurement at the full resolution of the DMD, which lasts only a fraction of a second, is possible for sparse images or in a situation when the field of view is limited but is a priori unknown. We propose the sampling and reconstruction strategies that enable us to reconstruct sparse images at the resolution of $1024 \times 768$ within the time of $0.3~$s. Non-sparse images are reconstructed with less details. The compression ratio is on the order of $0.4 \%$ which corresponds to an acquisition frequency of $7~$Hz. Sampling is differential, binary, and non-adaptive, and includes information on multiple partitioning of the image which later allows us to determine the actual field of view. Reconstruction is based on the differential Fourier domain regularized inversion (D-FDRI). The proposed SPI framework is an alternative to both adaptive SPI, which is challenging to implement in real time, and to classical compressive sensing image recovery methods, which are very slow at high resolutions.

arXiv:2206.01255

Compressive Fourier collocation methods for high-dimensional diffusion equations with periodic boundary conditions

High-dimensional Partial Differential Equations (PDEs) are a popular mathematical modelling tool, with applications ranging from finance to computational chemistry. However, standard numerical techniques for solving these PDEs are typically affected by the curse of dimensionality. In this work, we tackle this challenge while focusing on stationary diffusion equations defined over a high-dimensional domain with periodic boundary conditions. Inspired by recent progress in sparse function approximation in high dimensions, we propose a new method called compressive Fourier collocation. Combining ideas from compressive sensing and spectral collocation, our method replaces the use of structured collocation grids with Monte Carlo sampling and employs sparse recovery techniques, such as orthogonal matching pursuit and $\ell^1$ minimization, to approximate the Fourier coefficients of the PDE solution. We conduct a rigorous theoretical analysis showing that the approximation error of the proposed method is comparable with the best $s$-term approximation (with respect to the Fourier basis) to the solution. Using the recently introduced framework of random sampling in bounded Riesz systems, our analysis shows that the compressive Fourier collocation method mitigates the curse of dimensionality with respect to the number of collocation points under sufficient conditions on the regularity of the diffusion coefficient. We also present numerical experiments that illustrate the accuracy and stability of the method for the approximation of sparse and compressible solutions.

arXiv:1007.3753

Fast L1-Minimization Algorithms For Robust Face Recognition

L1-minimization refers to finding the minimum L1-norm solution to an underdetermined linear system b=Ax. Under certain conditions as described in compressive sensing theory, the minimum L1-norm solution is also the sparsest solution. In this paper, our study addresses the speed and scalability of its algorithms. In particular, we focus on the numerical implementation of a sparsity-based classification framework in robust face recognition, where sparse representation is sought to recover human identities from very high-dimensional facial images that may be corrupted by illumination, facial disguise, and pose variation. Although the underlying numerical problem is a linear program, traditional algorithms are known to suffer poor scalability for large-scale applications. We investigate a new solution based on a classical convex optimization framework, known as Augmented Lagrangian Methods (ALM). The new convex solvers provide a viable solution to real-world, time-critical applications such as face recognition. We conduct extensive experiments to validate and compare the performance of the ALM algorithms against several popular L1-minimization solvers, including interior-point method, Homotopy, FISTA, SESOP-PCD, approximate message passing (AMP) and TFOCS. To aid peer evaluation, the code for all the algorithms has been made publicly available.

arXiv:2205.10102공식 코드 · Papers with Code

Degradation-Aware Unfolding Half-Shuffle Transformer for Spectral Compressive Imaging

In coded aperture snapshot spectral compressive imaging (CASSI) systems, hyperspectral image (HSI) reconstruction methods are employed to recover the spatial-spectral signal from a compressed measurement. Among these algorithms, deep unfolding methods demonstrate promising performance but suffer from two issues. Firstly, they do not estimate the degradation patterns and ill-posedness degree from the highly related CASSI to guide the iterative learning. Secondly, they are mainly CNN-based, showing limitations in capturing long-range dependencies. In this paper, we propose a principled Degradation-Aware Unfolding Framework (DAUF) that estimates parameters from the compressed image and physical mask, and then uses these parameters to control each iteration. Moreover, we customize a novel Half-Shuffle Transformer (HST) that simultaneously captures local contents and non-local dependencies. By plugging HST into DAUF, we establish the first Transformer-based deep unfolding method, Degradation-Aware Unfolding Half-Shuffle Transformer (DAUHST), for HSI reconstruction. Experiments show that DAUHST significantly surpasses state-of-the-art methods while requiring cheaper computational and memory costs. Code and models will be released at https://github.com/caiyuanhao1998/MST

arXiv:2205.09708

Evaluation of the Effects of Compressive Spectrum Sensing Parameters on Primary User Behavior Estimation

As the Internet of Things (IoT) technology is being deployed, the demand for radio spectrum is increasing. Cognitive radio (CR) is one of the most promising solutions to allow opportunistic spectrum access for IoT secondary users through utilizing spectrum holes resulting from the underutilization of frequency spectrum. A CR needs to frequently sense the spectrum to avoid interference with primary users (PUs). Compressive spectrum sensing techniques have been gaining increasing interest in wideband spectrum sensing, as they reduce the need for high-rate analog-to-digital converters, reducing the complexity and energy requirements of the CR. In order to enhance spectrum sensing performance, researchers proposed to incorporate PU spectrum usage information into the process of spectrum sensing. Spectrum usage information can be obtained through pilot signals, geo-locational databases or through evaluation of previous spectrum sensing results. In this paper, we are studying the effects of compressive sensing parameters namely compression ratio, sensing period, and sensing duration on the estimation of primary user behavior statistics. We achieved an accurate estimation of the primary user's behavior while saving 40% of the sampling rate by using compressive spectrum sensing compared to traditional spectrum sensing with Nyquist rate sampling.

arXiv:2205.07185

Nonconvex ${L_ {1/2}} $-Regularized Nonlocal Self-similarity Denoiser for Compressive Sensing based CT Reconstruction

Compressive sensing (CS) based computed tomography (CT) image reconstruction aims at reducing the radiation risk through sparse-view projection data. It is usually challenging to achieve satisfying image quality from incomplete projections. Recently, the nonconvex ${{L_ {{1/2}}}} $-norm has achieved promising performance in sparse recovery, while the applications on imaging are unsatisfactory due to its nonconvexity. In this paper, we develop a ${{L_ {{1/2}}}} $-regularized nonlocal self-similarity (NSS) denoiser for CT reconstruction problem, which integrates low-rank approximation with group sparse coding (GSC) framework. Concretely, we first split the CT reconstruction problem into two subproblems, and then improve the CT image quality furtherly using our ${{L_ {{1/2}}}} $-regularized NSS denoiser. Instead of optimizing the nonconvex problem under the perspective of GSC, we particularly reconstruct CT image via low-rank minimization based on two simple yet essential schemes, which build the equivalent relationship between GSC based denoiser and low-rank minimization. Furtherly, the weighted singular value thresholding (WSVT) operator is utilized to optimize the resulting nonconvex ${{L_ {{1/2}}}} $ minimization problem. Following this, our proposed denoiser is integrated with the CT reconstruction problem by alternating direction method of multipliers (ADMM) framework. Extensive experimental results on typical clinical CT images have demonstrated that our approach can further achieve better performance than popular approaches.

arXiv:2205.07050

DECONET: an Unfolding Network for Analysis-based Compressed Sensing with Generalization Error Bounds

We present a new deep unfolding network for analysis-sparsity-based Compressed Sensing. The proposed network coined Decoding Network (DECONET) jointly learns a decoder that reconstructs vectors from their incomplete, noisy measurements and a redundant sparsifying analysis operator, which is shared across the layers of DECONET. Moreover, we formulate the hypothesis class of DECONET and estimate its associated Rademacher complexity. Then, we use this estimate to deliver meaningful upper bounds for the generalization error of DECONET. Finally, the validity of our theoretical results is assessed and comparisons to state-of-the-art unfolding networks are made, on both synthetic and real-world datasets. Experimental results indicate that our proposed network outperforms the baselines, consistently for all datasets, and its behaviour complies with our theoretical findings.

arXiv:2204.13599

Signal Recovery with Non-Expansive Generative Network Priors

We study compressive sensing with a deep generative network prior. Initial theoretical guarantees for efficient recovery from compressed linear measurements have been developed for signals in the range of a ReLU network with Gaussian weights and logarithmic expansivity: that is when each layer is larger than the previous one by a logarithmic factor. It was later shown that constant expansivity is sufficient for recovery. It has remained open whether the expansivity can be relaxed, allowing for networks with contractive layers (as often the case of real generators). In this work we answer this question, proving that a signal in the range of a Gaussian generative network can be recovered from few linear measurements provided that the width of the layers is proportional to the input layer size (up to log factors). This condition allows the generative network to have contractive layers. Our result is based on showing that Gaussian matrices satisfy a matrix concentration inequality which we term Range Restricted Weight Distribution Condition (R2WDC) and that weakens the Weight Distribution Condition (WDC) upon which previous theoretical guarantees were based. The WDC has also been used to analyze other signal recovery problems with generative network priors. By replacing the WDC with the R2WDC, we are able to extend previous results for signal recovery with expansive generative network priors to non-expansive ones. We discuss these extensions for phase retrieval, denoising, and spiked matrix recovery.

arXiv:2204.11640공식 코드 · Papers with Code

Hybrid ISTA: Unfolding ISTA With Convergence Guarantees Using Free-Form Deep Neural Networks

It is promising to solve linear inverse problems by unfolding iterative algorithms (e.g., iterative shrinkage thresholding algorithm (ISTA)) as deep neural networks (DNNs) with learnable parameters. However, existing ISTA-based unfolded algorithms restrict the network architectures for iterative updates with the partial weight coupling structure to guarantee convergence. In this paper, we propose hybrid ISTA to unfold ISTA with both pre-computed and learned parameters by incorporating free-form DNNs (i.e., DNNs with arbitrary feasible and reasonable network architectures), while ensuring theoretical convergence. We first develop HCISTA to improve the efficiency and flexibility of classical ISTA (with pre-computed parameters) without compromising the convergence rate in theory. Furthermore, the DNN-based hybrid algorithm is generalized to popular variants of learned ISTA, dubbed HLISTA, to enable a free architecture of learned parameters with a guarantee of linear convergence. To our best knowledge, this paper is the first to provide a convergence-provable framework that enables free-form DNNs in ISTA-based unfolded algorithms. This framework is general to endow arbitrary DNNs for solving linear inverse problems with convergence guarantees. Extensive experiments demonstrate that hybrid ISTA can reduce the reconstruction error with an improved convergence rate in the tasks of sparse recovery and compressive sensing.

arXiv:2204.07204공식 코드 · Papers with Code

Test-Time Training Can Close the Natural Distribution Shift Performance Gap in Deep Learning Based Compressed Sensing

Deep learning based image reconstruction methods outperform traditional methods. However, neural networks suffer from a performance drop when applied to images from a different distribution than the training images. For example, a model trained for reconstructing knees in accelerated magnetic resonance imaging (MRI) does not reconstruct brains well, even though the same network trained on brains reconstructs brains perfectly well. Thus there is a distribution shift performance gap for a given neural network, defined as the difference in performance when training on a distribution $P$ and training on another distribution $Q$, and evaluating both models on $Q$. In this work, we propose a domain adaptation method for deep learning based compressive sensing that relies on self-supervision during training paired with test-time training at inference. We show that for four natural distribution shifts, this method essentially closes the distribution shift performance gap for state-of-the-art architectures for accelerated MRI.

arXiv:2204.05503공식 코드 · Papers with Code

FSOINet: Feature-Space Optimization-Inspired Network for Image Compressive Sensing

In recent years, deep learning-based image compressive sensing (ICS) methods have achieved brilliant success. Many optimization-inspired networks have been proposed to bring the insights of optimization algorithms into the network structure design and have achieved excellent reconstruction quality with low computational complexity. But they keep the information flow in pixel space as traditional algorithms by updating and transferring the image in pixel space, which does not fully use the information in the image features. In this paper, we propose the idea of achieving information flow phase by phase in feature space and design a Feature-Space Optimization-Inspired Network (dubbed FSOINet) to implement it by mapping both steps of proximal gradient descent algorithm from pixel space to feature space. Moreover, the sampling matrix is learned end-to-end with other network parameters. Experiments show that the proposed FSOINet outperforms the existing state-of-the-art methods by a large margin both quantitatively and qualitatively. The source code is available on https://github.com/cwjjun/FSOINet.

arXiv:2204.05086

Blind Orthogonal Least Squares based Compressive Spectrum Sensing

As an enabling technique of cognitive radio (CR), compressive spectrum sensing (CSS) based on compressive sensing (CS) can detect the spectrum opportunities from wide frequency bands efficiently and accurately by using sub-Nyquist sampling rate. However, the sensing performance of most existing CSS excessively relies on the prior information such as spectrum sparsity or noise variance. Thus, a key challenge in practical CSS is how to work effectively even in the absence of such information. In this paper, we propose a blind orthogonal least squares based CSS algorithm (B-OLS-CSS), which functions properly without the requirement of prior information. Specifically, we develop a novel blind stopping rule for the OLS algorithm based on its probabilistic recovery condition. This innovative rule gets rid of the need of the spectrum sparsity or noise information, but only requires the computational-feasible mutual incoherence property of the given measurement matrix. Our theoretical analysis indicates that the signal-to-noise ratio required by the proposed B-OLS-CSS for achieving a certain sensing accuracy is relaxed than that by the benchmark CSS using the OMP algorithm, which is verified by extensive simulation results.

arXiv:2204.04963

A General Compressive Sensing Construct using Density Evolution

This paper proposes a general framework to design a sparse sensing matrix $\ensuremath{\mathbf{A}}\in \mathbb{R}^{m\times n}$, in a linear measurement system $\ensuremath{\mathbf{y}} = \ensuremath{\mathbf{Ax}}^{\natural} + \ensuremath{\mathbf{w}}$, where $\ensuremath{\mathbf{y}} \in \mathbb{R}^m$, $\ensuremath{\mathbf{x}}^{\natural}\in \RR^n$, and $\ensuremath{\mathbf{w}}$ denote the measurements, the signal with certain structures, and the measurement noise, respectively. By viewing the signal reconstruction from the measurements as a message passing algorithm over a graphical model, we leverage tools from coding theory in the design of low density parity check codes, namely the density evolution, and provide a framework for the design of matrix $\ensuremath{\mathbf{A}}$. Particularly, compared to the previous methods, our proposed framework enjoys the following desirable properties: ($i$) Universality: the design supports both regular sensing and preferential sensing, and incorporates them in a single framework; ($ii$) Flexibility: the framework can easily adapt the design of $\bA$ to a signal $\ensuremath{\mathbf{x}}^{\natural}$ with different underlying structures. As an illustration, we consider the $\ell_1$ regularizer, which correspond to Lasso, for both the regular sensing and preferential sensing scheme. Noteworthy, our framework can reproduce the classical result of Lasso, i.e., $m\geq c_0 k\log(n/k)$ (the regular sensing) with regular design after proper distribution approximation, where $c_0 > 0$ is some fixed constant. We also provide numerical experiments to confirm the analytical results and demonstrate the superiority of our framework whenever a preferential treatment of a sub-block of vector $\bx^{\natural}$ is required.

arXiv:2203.11278

One-Bit Compressive Sensing: Can We Go Deep and Blind?

One-bit compressive sensing is concerned with the accurate recovery of an underlying sparse signal of interest from its one-bit noisy measurements. The conventional signal recovery approaches for this problem are mainly developed based on the assumption that an exact knowledge of the sensing matrix is available. In this work, however, we present a novel data-driven and model-based methodology that achieves blind recovery; i.e., signal recovery without requiring the knowledge of the sensing matrix. To this end, we make use of the deep unfolding technique and develop a model-driven deep neural architecture which is designed for this specific task. The proposed deep architecture is able to learn an alternative sensing matrix by taking advantage of the underlying unfolded algorithm such that the resulting learned recovery algorithm can accurately and quickly (in terms of the number of iterations) recover the underlying compressed signal of interest from its one-bit noisy measurements. In addition, due to the incorporation of the domain knowledge and the mathematical model of the system into the proposed deep architecture, the resulting network benefits from enhanced interpretability, has a very small number of trainable parameters, and requires very small number of training samples, as compared to the commonly used black-box deep neural network alternatives for the problem at hand.