Person search aims to jointly perform person detection and re-identification by localizing and identifying a query person within a gallery of uncropped scene images. Existing methods predominantly utilize ImageNet pre-trained backbones, which may be suboptimal for capturing the complex spatial context and fine-grained identity cues necessary for person search. Moreover, they rely on a shared backbone feature for both person detection and re-identification, leading to suboptimal features due to conflicting optimization objectives. In this paper, we propose DiffPS (Diffusion Prior Knowledge for Person Search), a novel framework that leverages a pre-trained diffusion model while eliminating the optimization conflict between two sub-tasks. We analyze key properties of diffusion priors and propose three specialized modules: (i) Diffusion-Guided Region Proposal Network (DGRPN) for enhanced person localization, (ii) Multi-Scale Frequency Refinement Network (MSFRN) to mitigate shape bias, and (iii) Semantic-Adaptive Feature Aggregation Network (SFAN) to leverage text-aligned diffusion features. DiffPS sets a new state-of-the-art on CUHK-SYSU and PRW.

Unpaired point cloud completion is crucial for real-world applications, where ground-truth data for complete point clouds are often unavailable. By learning a completion map from unpaired incomplete and complete point cloud data, this task avoids the reliance on paired datasets. In this paper, we propose the \textit{Unbalanced Optimal Transport Map for Unpaired Point Cloud Completion (\textbf{UOT-UPC})} model, which formulates the unpaired completion task as the (Unbalanced) Optimal Transport (OT) problem. Our method employs a Neural OT model learning the UOT map using neural networks. Our model is the first attempt to leverage UOT for unpaired point cloud completion, achieving competitive or superior performance on both single-category and multi-category benchmarks. In particular, our approach is especially robust under the class imbalance problem, which is frequently encountered in real-world unpaired point cloud completion scenarios.

Multivariate long-term time series forecasting is critical for applications such as weather prediction, and traffic analysis. In addition, the implementation of Transformer variants has improved prediction accuracy. Following these variants, different input data process approaches also enhanced the field, such as tokenization techniques including point-wise, channel-wise, and patch-wise tokenization. However, previous studies still have limitations in time complexity, computational resources, and cross-dimensional interactions. To address these limitations, we introduce a novel CNN Autoencoder-based Score Attention mechanism (CASA), which can be introduced in diverse Transformers model-agnosticically by reducing memory and leading to improvement in model performance. Experiments on eight real-world datasets validate that CASA decreases computational resources by up to 77.7%, accelerates inference by 44.0%, and achieves state-of-the-art performance, ranking first in 87.5% of evaluated metrics. Our code is available at https://github.com/lmh9507/CASA.

Most existing image captioning evaluation metrics focus on assigning a single numerical score to a caption by comparing it with reference captions.However, these methods do not provide an explanation for the assigned score.Moreover, reference captions are expensive to acquire.In this paper, we propose FLEUR 1 , an explainable reference-free metric to introduce explainability into image captioning evaluation metrics.By leveraging a large multimodal model, FLEUR can evaluate the caption against the image without the need for reference captions, and provide the explanation for the assigned score.We introduce score smoothing to align as closely as possible with human judgment and to be robust to user-defined grading criteria.FLEUR achieves high correlations with human judgment across various image captioning evaluation benchmarks and reaches state-of-the-art results on Flickr8k-CF, COMPOSITE, and Pascal-50S within the domain of reference-free evaluation metrics.Our source code and results are publicly available at: https://github.com/ Yebin46/FLEUR. * Equal contribution. Correspondence to: Myungjoo Kang 1We choose a word in French that means 'flower', in line with other French-named evaluation metrics.2 A reference caption refers to the human-annotated caption for an image.A candidate caption refers to the caption that is to be evaluated. Score: 0.02418The caption does not accurately describe the image.The image shows a boy walking out of the ocean water on the beach, holding a paddle.There is no dog or any object in the dog's mouth in the image.The caption is unrelated to the actual content of the image, which is why it receives a low score.A black dog is coming out of the ocean water on the beach with something in its mouth.BLEU-4: 0.25 ROUGE-L: 0.

Optimal Transport (OT) problem aims to find a transport plan that bridges two distributions while minimizing a given cost function. OT theory has been widely utilized in generative modeling. In the beginning, OT distance has been used as a measure for assessing the distance between data and generated distributions. Recently, OT transport map between data and prior distributions has been utilized as a generative model. These OT-based generative models share a similar adversarial training objective. In this paper, we begin by unifying these OT-based adversarial methods within a single framework. Then, we elucidate the role of each component in training dynamics through a comprehensive analysis of this unified framework. Moreover, we suggest a simple but novel method that improves the previously best-performing OT-based model. Intuitively, our approach conducts a gradual refinement of the generated distribution, progressively aligning it with the data distribution. Our approach achieves a FID score of 2.51 on CIFAR-10 and 5.99 on CelebA-HQ-256, outperforming unified OT-based adversarial approaches.

While backpropagation (BP) has achieved widespread success in deep learning, it faces two prominent challenges: computational inefficiency and biological implausibility. In response to these challenges, local supervision, encompassing Local Learning (LL) and Forward Learning (FL), has emerged as a promising research direction. LL employs module-wise BP to achieve competitive results yet relies on module-wise auxiliary networks, which increase memory and parameter demands. Conversely, FL updates layer weights without BP and auxiliary networks but falls short of BP’s performance. This paper proposes a simple yet effective objective within a contrastive learning framework for local supervision without auxiliary networks. Given the insight that the existing contrastive learning framework for local supervision is susceptible to task-irrelevant information without auxiliary networks, we present DICTIONARY CONTRASTIVE LEARNING (DCL) that optimizes the similarity between local features and label embeddings. Our method using static label embeddings yields substantial performance improvements in the FL scenario, outperforming state-of-the-art FL approaches. Moreover, our method using adaptive label embeddings closely approaches the performance achieved by LL while achieving superior memory and parameter efficiency.

We propose Feature-aligned N-BEATS as a domain-generalized time series forecasting model. It is a nontrivial extension of N-BEATS with doubly residual stacking principle (Oreshkin et al. [45]) into a representation learning framework. In particular, it revolves around marginal feature probability measures induced by the intricate composition of residual and feature extracting operators of N-BEATS in each stack and aligns them stack-wise via an approximate of an optimal transport distance referred to as the Sinkhorn divergence. The training loss consists of an empirical risk minimization from multiple source domains, i.e., forecasting loss, and an alignment loss calculated with the Sinkhorn divergence, which allows the model to learn invariant features stack-wise across multiple source data sequences while retaining N-BEATS’s interpretable design and forecasting power. Comprehensive experimental evaluations with ablation studies are provided and the corresponding results demonstrate the proposed model’s forecasting and generalization capabilities.

Wasserstein gradient flow (WGF) describes the gradient dynamics of probability density within the Wasserstein space. WGF provides a promising approach for conducting optimization over the probability distributions. Numerically approximating the continuous WGF requires the time discretization method. The most well-known method for this is the JKO scheme. In this regard, previous WGF models employ the JKO scheme and parametrized transport map for each JKO step. However, this approach results in quadratic training complexity $O(K^2)$ with the number of JKO step $K$. This severely limits the scalability of WGF models. In this paper, we introduce a scalable WGF-based generative model, called Semi-dual JKO (S-JKO). Our model is based on the semi-dual form of the JKO step, derived from the equivalence between the JKO step and the Unbalanced Optimal Transport. Our approach reduces the training complexity to $O(K)$. We demonstrate that our model significantly outperforms existing WGF-based generative models, achieving FID scores of 2.62 on CIFAR-10 and 6.42 on CelebA-HQ-256, which are comparable to state-of-the-art image generative models.

This paper analyzes the inverse relationship between the order of partial differential equations (PDEs) and the convergence of gradient descent in physics-informed neural networks (PINNs) with the power of ReLU activation. The integration of the PDE into a loss function endows PINNs with a distinctive feature to require computing derivatives of model up to the PDE order. Although it has been empirically observed that PINNs encounter difficulties in convergence when dealing with high-order or high-dimensional PDEs, a comprehensive theoretical understanding of this issue remains elusive. This paper offers theoretical support for this pathological behavior by demonstrating that the gradient flow converges in a lower probability when the PDE order is higher. In addition, we show that PINNs struggle to address high-dimensional problems because the influence of dimensionality on convergence is exacerbated with increasing PDE order. To address the pathology, we use the insights garnered to consider variable splitting that decomposes the high-order PDE into a system of lower-order PDEs. We prove that by reducing the differential order, the gradient flow of variable splitting is more likely to converge to the global optimum. Furthermore, we present numerical experiments in support of our theoretical claims.

Abstract Recently, several types of neural operators have been developed, including deep operator networks, graph neural operators, and Multiwavelet-based operators. Compared with these models, the Fourier neural operator (FNO), a physics-inspired machine learning method, is computationally efficient and can learn nonlinear operators between function spaces independent of a certain finite basis. This study investigated the bounding of the Rademacher complexity of the FNO based on specific group norms. Using capacity based on these norms, we bound the generalization error of the model. In addition, we investigate the correlation between the empirical generalization error and the proposed capacity of FNO. We infer that the type of group norm determines the information about the weights and architecture of the FNO model stored in capacity. The experimental results offer insight into the impact of the number of modes used in the FNO model on the generalization error. The results confirm that our capacity is an effective index for estimating generalization errors.

Recent studies have substantiated that machine learning algorithms including convolutional neural networks often suffer from unreliable generalizations when there is a significant gap between the source and target data distributions. To mitigate this issue, a predetermined distribution shift has been addressed independently (e.g., single domain generalization, de-biasing). However, a distribution mismatch cannot be clearly estimated because the target distribution is unknown at training. Therefore, a conservative approach robust on unexpected diverse distributions is more desirable in practice. Our work starts from a motivation to allow adaptive inference once we know the target, since it is accessible only at testing. Instead of assuming and fixing the target distribution at training, our proposed approach allows adjusting the feature space the model refers to at every prediction, i.e., instance-wise adaptive inference. The extensive evaluation demonstrates our method is effective for generalization on diverse distributions.

Denoising diffusion models (DDMs) have recently attracted increasing attention by showing impressive synthesis quality. DDMs are built on a diffusion process that pushes data to the noise distribution and the models learn to denoise. In this paper, we establish the interpretation of DDMs in terms of image restoration (IR). Integrating IR literature allows us to use an alternative objective and diverse forward processes, not confining to the diffusion process. By imposing prior knowledge on the loss function grounded on MAP-based estimation, we eliminate the need for the expensive sampling of DDMs. Also, we propose a multi-scale training, which improves the performance compared to the diffusion process, by taking advantage of the flexibility of the forward process. Experimental results demonstrate that our model improves the quality and efficiency of both training and inference. Furthermore, we show the applicability of our model to inverse problems. We believe that our framework paves the way for designing a new type of flexible general generative model.

Combinatorial generalization refers to the ability to collect and assemble various attributes from diverse data to generate novel unexperienced data. This ability is considered a necessary passing point for achieving human-level intelligence. To achieve this ability, previous unsupervised approaches mainly focused on learning the disentangled representation, such as the variational autoencoder. However, recent studies discovered that the disentangled representation is insufficient for combinatorial generalization and is not even correlated. In this regard, we propose a novel framework for data generation that can robustly generalize under these distribution shift situations. Instead of representing each data, our model discovers the fundamental transformation between a pair of data by simulating a group action. To test the combinatorial generalizability, we evaluated our model in two settings: Recombination-to-Element and Recombination-to-Range. The experiments demonstrated that our method has quantitatively and qualitatively superior generalizability and generates better images than traditional models.

Optimal Transport (OT) problem investigates a transport map that bridges two distributions while minimizing a given cost function. In this regard, OT between tractable prior distribution and data has been utilized for generative modeling tasks. However, OT-based methods are susceptible to outliers and face optimization challenges during training. In this paper, we propose a novel generative model based on the semi-dual formulation of Unbalanced Optimal Transport (UOT). Unlike OT, UOT relaxes the hard constraint on distribution matching. This approach provides better robustness against outliers, stability during training, and faster convergence. We validate these properties empirically through experiments. Moreover, we study the theoretical upper-bound of divergence between distributions in UOT. Our model outperforms existing OT-based generative models, achieving FID scores of 2.97 on CIFAR-10 and 6.36 on CelebA-HQ-256. The code is available at \url{https://github.com/Jae-Moo/UOTM}.

The aim of this paper is the reconstruction of a smooth surface from an unorganized point cloud sampled by a closed surface, with the preservation of geometric shapes, without any further information other than the point cloud. Implicit neural representations (INRs) have recently emerged as a promising approach to surface reconstruction. However, the reconstruction quality of existing methods relies on ground truth implicit function values or surface normal vectors. In this paper, we show that proper supervision of partial differential equations and fundamental properties of differential vector fields are sufficient to robustly reconstruct high-quality surfaces. We cast the $p$-Poisson equation to learn a signed distance function (SDF) and the reconstructed surface is implicitly represented by the zero-level set of the SDF. For efficient training, we develop a variable splitting structure by introducing a gradient of the SDF as an auxiliary variable and impose the $p$-Poisson equation directly on the auxiliary variable as a hard constraint. Based on the curl-free property of the gradient field, we impose a curl-free constraint on the auxiliary variable, which leads to a more faithful reconstruction. Experiments on standard benchmark datasets show that the proposed INR provides a superior and robust reconstruction. The code is available at https://github.com/Yebbi/PINC.

A recurrent neural network (RNN) is a widely used deep-learning network for dealing with sequential data. Imitating a dynamical system, an infinite-width RNN can approximate any open dynamical system in a compact domain. In general, deep narrow networks with bounded width and arbitrary depth are more effective than wide shallow networks with arbitrary width and bounded depth in practice; however, the universal approximation theorem for deep narrow structures has yet to be extensively studied. In this study, we prove the universality of deep narrow RNNs and show that the upper bound of the minimum width for universality can be independent of the length of the data. Specifically, we show a deep RNN with ReLU activation can approximate any continuous function or $L^p$ function with the widths $d_x+d_y+3$ and $\max\{d_x+1,d_y\}$, respectively, where the target function maps a finite sequence of vectors in $\mathbb{R}^{d_x}$ to a finite sequence of vectors in $\mathbb{R}^{d_y}$. We also compute the additional width required if the activation function is sigmoid or more. In addition, we prove the universality of other recurrent networks, such as bidirectional RNNs. Bridging a multi-layer perceptron and an RNN, our theory and technique can shed light on further research on deep RNNs.

Although convolutional neural networks (CNNs) achieve state-of-the-art in image classification, recent works address their unreliable predictions due to their excessive dependence on biased training data. Existing unbiased modeling postulates that the bias in the dataset is obvious to know, but it is actually unsuited for image datasets including countless sensory attributes. To mitigate this issue, we present a new scenario that does not necessitate a predefined bias. Under the observation that CNNs do have multi-variant and unbiased representations in the model, we propose a conservative framework that employs this internal information for unbiased learning. Specifically, this mechanism is implemented via hierarchical features captured along the multiple layers and orthogonal regularization. Extensive evaluations on public benchmarks demonstrate our method is effective for unbiased learning.11Source code: https://github.com/aandyjeon/UBNet

Image processing via convolutional neural network (CNN) has been developed rapidly for remote sensing technology. Moreover, techniques for accurately extracting building footprints from remote sensed images have attracted considerable interest owing to their wide variety of common applications, including monitoring natural disasters and urban development. Extraction of building footprints can be performed easily by semantic segmentation using U-Net-like CNN architectures. However, obtaining precise boundaries of segmentation masks remains challenging due to various impediments surrounding target objects. In this study, we propose a method to elaborate edges of buildings detected in remote sensed images to enhance the boundaries of segmentation masks. The proposed method adoptsholistically nested edge detection (HED), which extracts edge features at an encoder of a given architecture. In the proposedboundary enhancement (BE) module, an extracted edge and segmentation mask are combined, sharing mutual information. To enable the proposed method efficiently to adapt to a wide variety of conditions, we design a distinctive approach adopting a HED unit and BE module, which is applicable to various semantic segmentation networks containing encoder-decoder structures. Experiments were conducted on five different datasets (DeepGlobe, Urban3D, WHU [high-resolution (HR), low-resolution (LR)], and Massachusetts). The results demonstrate that our proposed approaches improved on the performance of prior methods for extracting building footprints. Comparative experiments were conducted on various backbone architectures including U-Net, ResUNet++, TernausNet, and U-shape spatial pyramid pooling (USPP) to ensure the effectiveness of the proposed method. Based on various evaluation metrics and qualitative analysis, our results show that the proposed method achieved improved performance compared with prior methods for all datasets and backbone networks.

We present a novel, blind single image deblurring method that utilizes information regarding blur kernels. Our model solves the deblurring problem by dividing it into two successive tasks: (1) blur kernel estimation and (2) sharp image restoration. We first introduce a kernel estimation network that produces adaptive blur kernels based on the analysis of the blurred image. The network learns the blur pattern of the input image and trains to generate the estimation of image-specific blur kernels. Subsequently, we propose a deblurring network that restores sharp images using the estimated blur kernel. To use the kernel efficiently, we propose a kernel-adaptive AE block to apply the kernel information on the feature. We evaluate our model on REDS, GOPRO and Flickr2K datasets using various Gaussian blur kernels. Experiments show that our model can achieve state-of-the-art results on each dataset.

This article introduces a variational color image decomposition and restoration model. The aim is to recover an image from its degraded version, while simultaneously decomposing the image into its cartoon and texture components. The energy involves adaptive higher-order regularizers, incorporated with an edge indicator function. This not only helps cartoon and texture decomposition, but also provides higher quality image restoration by ameliorating the staircasing effect that arises in total variation regularization methods. To realize the proposed models, we present fast and efficient iterative algorithms based on a variable splitting scheme and an augmented Lagrangian method. A convergence analysis of the proposed algorithms is also presented under certain conditions. Numerical results and comparisons demonstrate that the proposed model is more effective than state-of-the-art methods for both image decomposition and restoration.

The fully developed speckle (multiplicative noise) naturally appears in coherent imaging systems, such as synthetic aperture radar. Since the speckle is multiplicative, it is difficult to interpret observed data. Total variation (TV) based variational models have recently been used in the removal of the speckle because of the strong edge preserving property of TV and reasonable computational cost. However, the fidelity term (or negative log-likelihood) of the original variational model [G. Aubert and J.-F. Aujol, SIAM J. Appl. Math., 68 (2008), pp. 925--946], which appears on maximum a posteriori (MAP) estimation, is not convex. Recently, the logarithmic transformation and the $m$th root transformation have been proposed to relax the nonconvexity. It is empirically observed that the $m$th root transform based variational model outperforms the log transform based variational model. However, the performance of the $m$th root transform based model critically depends on the choice of $m$. In this paper, we propose the two-level convex relaxed variational model; i.e., we relax the original variational model by using the $m$th root transformation and the concave conjugate. We also adapt the two-block nonlinear Gauss--Seidel method to solve the proposed model. The performance of the proposed model does not depend on the choice of $m$, and the model shows overall better performance than the logarithmic transformed variational model and the $m$th root transformed variational model.

In this paper, we propose a new method of segmenting an image into several sets of pixels with similar intensity values called regions. A multiple-region segmentation problem is unstable because the result considerably depends on the number of regions given a priori. Therefore, one of the most important tasks in solving the problem is automatically finding the number of regions. The method we propose is able to find the reasonable number of distinct regions not only for clean images but also for noisy ones. Our method is made up of two procedures. First, we develop the adaptive global maximum clustering. In this procedure, we deal with an image histogram and automatically obtain the number of significant local maxima of the histogram. This number indicates the number of different regions in the image. Second, we derive a simple and fast calculation to segment an image composed of distinct multiple regions. Then, we split an image into multiple regions according to the previous procedure. Finally, we show the efficiency of our method by comparing it with other previous methods.