SCSA: A Plug-and-Play Semantic Continuous-Sparse Attention for Arbitrary Semantic Style Transfer

Attention-based arbitrary style transfer methods, including CNN-based, Transformer-based, and Diffusion-based, have flourished and produced high-quality stylized images. However, they perform poorly on the content and style images with the same semantics, i.e., the style of the corresponding semantic region of the generated stylized image is inconsistent with that of the style image. We argue that the root cause lies in their failure to consider the relationship between local regions and semantic regions. To address this issue, we propose a plug-and-play semantic continuous-sparse attention, dubbed SCSA, for arbitrary semantic style transfer -- each query point considers certain key points in the corresponding semantic region. Specifically, semantic continuous attention ensures each query point fully attends to all the continuous key points in the same semantic region that reflect the overall style characteristics of that region; Semantic sparse attention allows each query point to focus on the most similar sparse key point in the same semantic region that exhibits the specific stylistic texture of that region. By combining the two modules, the resulting SCSA aligns the overall style of the corresponding semantic regions while transferring the vivid textures of these regions. Qualitative and quantitative results prove that SCSA enables attention-based arbitrary style transfer methods to produce high-quality semantic stylized images.

Improving Decoupled Posterior Sampling for Inverse Problems using Data Consistency Constraint

Diffusion models have shown strong performances in solving inverse problems through posterior sampling while they suffer from errors during earlier steps. To mitigate this issue, several Decoupled Posterior Sampling methods have been recently proposed. However, the reverse process in these methods ignores measurement information, leading to errors that impede effective optimization in subsequent steps. To solve this problem, we propose Guided Decoupled Posterior Sampling (GDPS) by integrating a data consistency constraint in the reverse process. The constraint performs a smoother transition within the optimization process, facilitating a more effective convergence toward the target distribution. Furthermore, we extend our method to latent diffusion models and Tweedie's formula, demonstrating its scalability. We evaluate GDPS on the FFHQ and ImageNet datasets across various linear and nonlinear tasks under both standard and challenging conditions. Experimental results demonstrate that GDPS achieves state-of-the-art performance, improving accuracy over existing methods.

Average case analysis of Lasso under ultra-sparse conditions

We analyze the performance of the least absolute shrinkage and selection operator (Lasso) for the linear model when the number of regressors $N$ grows larger keeping the true support size $d$ finite, i.e., the ultra-sparse case. The result is based on a novel treatment of the non-rigorous replica method in statistical physics, which has been applied only to problem settings where $N$ ,$d$ and the number of observations $M$ tend to infinity at the same rate. Our analysis makes it possible to assess the average performance of Lasso with Gaussian sensing matrices without assumptions on the scaling of $N$ and $M$, the noise distribution, and the profile of the true signal. Under mild conditions on the noise distribution, the analysis also offers a lower bound on the sample complexity necessary for partial and perfect support recovery when $M$ diverges as $M = O(\log N)$. The obtained bound for perfect support recovery is a generalization of that given in previous literature, which only considers the case of Gaussian noise and diverging $d$. Extensive numerical experiments strongly support our analysis.

QCM-SGM+: Improved Quantized Compressed Sensing With Score-Based Generative Models for General Sensing Matrices

In realistic compressed sensing (CS) scenarios, the obtained measurements usually have to be quantized to a finite number of bits before transmission and/or storage, thus posing a challenge in recovery, especially for extremely coarse quantization such as 1-bit sign measurements. Recently Meng & Kabashima proposed an efficient quantized compressed sensing algorithm called QCS-SGM using the score-based generative models as an implicit prior. Thanks to the power of score-based generative models in capturing the rich structure of the prior, QCS-SGM achieves remarkably better performances than previous quantized CS methods. However, QCS-SGM is restricted to (approximately) row-orthogonal sensing matrices since otherwise the likelihood score becomes intractable. To address this challenging problem, in this paper we propose an improved version of QCS-SGM, which we call QCS-SGM+, which also works well for general matrices. The key idea is a Bayesian inference perspective of the likelihood score computation, whereby an expectation propagation algorithm is proposed to approximately compute the likelihood score. Experiments on a variety of baseline datasets demonstrate that the proposed QCS-SGM+ outperforms QCS-SGM by a large margin when sensing matrices are far from row-orthogonal.

Diffusion Model Based Posterior Sampling for Noisy Linear Inverse Problems

We consider the ubiquitous linear inverse problems with additive Gaussian noise and propose an unsupervised general-purpose sampling approach called diffusion model based posterior sampling (DMPS) to reconstruct the unknown signal from noisy linear measurements. Specifically, the prior of the unknown signal is implicitly modeled by one pre-trained diffusion model (DM). In posterior sampling, to address the intractability of exact noise-perturbed likelihood score, a simple yet effective noise-perturbed pseudo-likelihood score is introduced under the uninformative prior assumption. While DMPS applies to any kind of DM with proper modifications, we focus on the ablated diffusion model (ADM) as one specific example and evaluate its efficacy on a variety of linear inverse problems such as image super-resolution, denoising, deblurring, colorization. Experimental results demonstrate that, for both in-distribution and out-of-distribution samples, DMPS achieves highly competitive or even better performances on various tasks while being 3 times faster than the leading competitor. The code to reproduce the results is available at https://github.com/mengxiangming/dmps.

A Unitary Transform Based Generalized Approximate Message Passing

We consider the problem of recovering an unknown signal ${\mathbf x}\in {\mathbb R}^n$ from general nonlinear measurements obtained through a generalized linear model (GLM), i.e., ${\mathbf y}= f\left({\mathbf A}{\mathbf x}+{\mathbf w}\right)$, where $f(\cdot)$ is a componentwise nonlinear function. Based on the unitary transform approximate message passing (UAMP) and expectation propagation, a unitary transform based generalized approximate message passing (GUAMP) algorithm is proposed for general measurement matrices $\bf{A}$, in particular highly correlated matrices. Experimental results on quantized compressed sensing demonstrate that the proposed GUAMP significantly outperforms state-of-the-art GAMP and GVAMP under correlated matrices $\bf{A}$.

Exact Solutions of a Deep Linear Network

This work finds the exact solutions to a deep linear network with weight decay and stochastic neurons, a fundamental model for understanding the landscape of neural networks. Our result implies that weight decay strongly interacts with the model architecture and can create bad minima in a network with more than $1$ hidden layer, qualitatively different for a network with only $1$ hidden layer. As an application, we also analyze stochastic nets and show that their prediction variance vanishes to zero as the stochasticity, the width, or the depth tends to infinity.

Stochastic Neural Networks with Infinite Width are Deterministic

This work theoretically studies stochastic neural networks, a main type of neural network in use. Specifically, we prove that as the width of an optimized stochastic neural network tends to infinity, its predictive variance on the training set decreases to zero. Two common examples that our theory applies to are neural networks with dropout and variational autoencoders. Our result helps better understand how stochasticity affects the learning of neural networks and thus design better architectures for practical problems.

On Model Selection Consistency of Lasso for High-Dimensional Ising Models on Tree-like Graphs

We consider the problem of high-dimensional Ising model selection using neighborhood-based least absolute shrinkage and selection operator (Lasso). It is rigorously proved that under some mild coherence conditions on the population covariance matrix of the Ising model, consistent model selection can be achieved with sample sizes $n=\Omega{(d^3\log{p})}$ for any tree-like graph in the paramagnetic phase, where $p$ is the number of variables and $d$ is the maximum node degree. When the same conditions are imposed directly on the sample covariance matrices, it is shown that a reduced sample size $n=\Omega{(d^2\log{p})}$ suffices. The obtained sufficient conditions for consistent model selection with Lasso are the same in the scaling of the sample complexity as that of $\ell_1$-regularized logistic regression. Given the popularity and efficiency of Lasso, our rigorous analysis provides a theoretical backing for its practical use in Ising model selection.

Ising Model Selection Using $\ell_{1}$-Regularized Linear Regression

We theoretically investigate the performance of $\ell_{1}$-regularized linear regression ($\ell_1$-LinR) for the problem of Ising model selection using the replica method from statistical mechanics. The regular random graph is considered under paramagnetic assumption. Our results show that despite model misspecification, the $\ell_1$-LinR estimator can successfully recover the graph structure of the Ising model with $N$ variables using $M=\mathcal{O}\left(\log N\right)$ samples, which is of the same order as that of $\ell_{1}$-regularized logistic regression. Moreover, we provide a computationally efficient method to accurately predict the non-asymptotic performance of the $\ell_1$-LinR estimator with moderate $M$ and $N$. Simulations show an excellent agreement between theoretical predictions and experimental results, which supports our findings.