arXiv subjects
Publications and source records attributed to Xiaoqi Ma.
This paper introduces a unified analysis framework of equal-order hybridized discontinuous finite element (HDG) methods. The general framework covers standard HDG, embedded discontinuous finite element, and embedded-hybridized discontinuous finite element methods.
We introduce LongCat-Image, a pioneering open-source and bilingual (Chinese-English) foundation model for image generation, designed to address core challenges in multilingual text rendering, photorealism, deployment efficiency, and developer accessibility prevalent in current leading models. 1) We achieve this through rigorous data curation strategies across the pre-training, mid-training, and SFT stages, complemented by the coordinated use of curated reward models during the RL phase. This strategy establishes the model as a new state-of-the-art (SOTA), delivering superior text-rendering capabilities and remarkable photorealism, and significantly enhancing aesthetic quality. 2) Notably, it sets a new industry standard for Chinese character rendering. By supporting even complex and rare characters, it outperforms both major open-source and commercial solutions in coverage, while also achieving superior accuracy. 3) The model achieves remarkable efficiency through its compact design. With a core diffusion model of only 6B parameters, it is significantly smaller than the nearly 20B or larger Mixture-of-Experts (MoE) architectures common in the field. This ensures minimal VRAM usage and rapid inference, significantly reducing deployment costs. Beyond generation, LongCat-Image also excels in image editing, achieving SOTA results on standard benchmarks with superior editing consistency compared to other open-source works. 4) To fully empower the community, we have established the most comprehensive open-source ecosystem to date. We are releasing not only multiple model versions for text-to-image and image editing, including checkpoints after mid-training and post-training stages, but also the entire toolchain of training procedure. We believe that the openness of LongCat-Image will provide robust support for developers and researchers, pushing the frontiers of visual content creation.
As the frequency and complexity of Distributed Denial-of-Service (DDoS) attacks continue to increase, the level of threats posed to Smart Internet of Things (SIoT) business environments have also increased. These environments generally have several interconnected SIoT systems and devices that are integral to daily operations, usually depending on cloud infrastructure and real-time data analytics, which require continuous availability and secure data exchange. Conventional detection mechanisms, while useful in static or traditional network environments, often are inadequate in responding to the needs of these dynamic and diverse SIoT networks. In this paper, we introduce a novel trust-based DDoS detection model tailored to meet the unique requirements of smart business environments. The proposed model incorporates a trust evaluation engine that continuously monitors node behaviour, calculating trust scores based on packet delivery ratio, response time, and anomaly detection. These trust metrics are then aggregated by a central trust-based repository that uses inherent trust values to identify traffic patterns indicative of DDoS attacks. By integrating both trust scores and central trust-based outputs, the trust calculation is enhanced, ensuring that threats are accurately identified and addressed in real-time. The model demonstrated a significant improvement in detection accuracy, and a low false-positive rate with enhanced scalability and adaptability under TCP SYN, Ping Flood, and UDP Flood attacks. The results show that a trust-based approach provides an effective, lightweight alternative for securing resource-constrained business IoT environments.
Text-to-image (T2I) generation has achieved remarkable progress in instruction following and aesthetics. However, a persistent challenge is the prevalence of physical artifacts, such as anatomical and structural flaws, which severely degrade perceptual quality and limit application. Given the diversity and complexity of these artifacts, a systematic and fine-grained evaluation framework is required, which is lacking in current benchmarks. To fill this gap, we introduce MagicMirror, a comprehensive framework for artifacts assessment. We first establish a detailed taxonomy of generated image artifacts. Guided by this taxonomy, we manually annotate MagicData340K, the first human-annotated large-scale dataset of 340K generated images with fine-grained artifact labels. Building on this dataset, we train MagicAssessor, a Vision-Language Model (VLM) that provides detailed assessments and corresponding labels. To overcome challenges like class imbalance and reward hacking, we design a novel data sampling strategy and a multi-level reward system for Group Relative Policy Optimization (GRPO). Finally, we leverage MagicAssessor to construct MagicBench, an automated benchmark for evaluating the image artifacts of current T2I models. Our evaluation with MagicBench reveals that despite their widespread adoption, even top-tier models like GPT-image-1 are consistently plagued by significant artifacts, highlighting artifact reduction as a critical frontier for future T2I development. Project page: https://wj-inf.github.io/MagicMirror-page/.
With deeper exploration of diffusion model, developments in the field of image generation have triggered a boom in image creation. As the quality of base-model generated images continues to improve, so does the demand for further application like image editing. In recent years, many remarkable works are realizing a wide variety of editing effects. However, the wide variety of editing types and diverse editing approaches have made it difficult for researchers to establish a comprehensive view of the development of this field. In this survey, we summarize the image editing field from four aspects: tasks definition, methods classification, results evaluation and editing datasets. First, we provide a definition of image editing, which in turn leads to a variety of editing task forms from the perspective of operation parts and manipulation actions. Subsequently, we categorize and summary methods for implementing editing into three categories: inversion-based, fine-tuning-based and adapter-based. In addition, we organize the currently used metrics, available datasets and corresponding construction methods. At the end, we present some visions for the future development of the image editing field based on the previous summaries.
Rapid advancements have been made in extending Large Language Models (LLMs) to Large Multi-modal Models (LMMs). However, extending input modality of LLMs to video data remains a challenging endeavor, especially for long videos. Due to insufficient access to large-scale high-quality video data and the excessive compression of visual features, current methods exhibit limitations in effectively processing long videos. In this paper, we introduce Kangaroo, a powerful Video LMM aimed at addressing these challenges. Confronted with issue of inadequate training data, we develop a data curation system to build a large-scale dataset with high-quality annotations for vision-language pre-training and instruction tuning. In addition, we design a curriculum training pipeline with gradually increasing resolution and number of input frames to accommodate long videos. Evaluation results demonstrate that, with 8B parameters, Kangaroo achieves state-of-the-art performance across a variety of video understanding benchmarks while exhibiting competitive results on others. Particularly, on benchmarks specialized for long videos, Kangaroo excels some larger models with over 10B parameters and proprietary models.
This paper presents the first analysis of parameter-uniform convergence for a hybridizable discontinuous Galerkin (HDG) method applied to a singularly perturbed convection-diffusion problem in 2D using a Shishkin mesh. The primary difficulty lies in accurately estimating the convection term in the layer, where existing methods often fall short. To address this, a novel error control technique is employed, along with reasonable assumptions regarding the stabilization function. The results show that, with polynomial degrees not exceeding $k$, the method achieves supercloseness of almost $k+\frac{1}{2}$ order in an energy norm. Numerical experiments confirm the theoretical accuracy and efficiency of the proposed method.
This paper investigates the supercloseness of a singularly perturbed convection diffusion problem using the direct discontinuous Galerkin (DDG) method on a Shishkin mesh. The main technical difficulties lie in controlling the diffusion term inside the layer, the convection term outside the layer, and the inter element jump term caused by the discontinuity of the numerical solution. The main idea is to design a new composite interpolation, in which a global projection is used outside the layer to satisfy the interface conditions determined by the selection of numerical flux, thereby eliminating or controlling the troublesome terms on the unit interface; and inside the layer, Gauß Lobatto projection is used to improve the convergence order of the diffusion term. On the basis of that, by selecting appropriate parameters in the numerical flux, we obtain the supercloseness result of almost $k+1$ order under an energy norm. Numerical experiments support our main theoretical conclusion.
Breast cancer has reached the highest incidence rate worldwide among all malignancies since 2020. Breast imaging plays a significant role in early diagnosis and intervention to improve the outcome of breast cancer patients. In the past decade, deep learning has shown remarkable progress in breast cancer imaging analysis, holding great promise in interpreting the rich information and complex context of breast imaging modalities. Considering the rapid improvement in deep learning technology and the increasing severity of breast cancer, it is critical to summarize past progress and identify future challenges to be addressed. This paper provides an extensive review of deep learning-based breast cancer imaging research, covering studies on mammogram, ultrasound, magnetic resonance imaging, and digital pathology images over the past decade. The major deep learning methods and applications on imaging-based screening, diagnosis, treatment response prediction, and prognosis are elaborated and discussed. Drawn from the findings of this survey, we present a comprehensive discussion of the challenges and potential avenues for future research in deep learning-based breast cancer imaging.
The continuous rise in the adoption of emerging technologies such as Internet of Things (IoT) by businesses has brought unprecedented opportunities for innovation and growth. However, due to the distinct characteristics of these emerging IoT technologies like real-time data processing, Self-configuration, interoperability, and scalability, they have also introduced some unique cybersecurity challenges, such as malware attacks, advanced persistent threats (APTs), DoS /DDoS (Denial of Service & Distributed Denial of Service attacks) and insider threats. As a result of these challenges, there is an increased need for improved cybersecurity approaches and efficient management solutions to ensure the privacy and security of communication within IoT networks. One proposed security approach is the utilization of trust-based systems and is the focus of this study. This research paper presents a systematic literature review on the Trust-based cybersecurity security approaches for IoT. A total of 23 articles were identified that satisfy the review criteria. We highlighted the common trust-based mitigation techniques in existence for dealing with these threats and grouped them into three major categories, namely: Observation-Based, Knowledge-Based & Cluster-Based systems. Finally, several open issues were highlighted, and future research directions presented.
This article investigates a local discontinuous Galerkin (LDG) method for one-dimensional and two-dimensional singularly perturbed reaction-diffusion problems on a Shishkin mesh. During this process, due to the inability of the energy norm to fully capture the behavior of the boundary layers appearing in the solutions, a balanced norm is introduced. By designing novel numerical fluxes and constructing special interpolations, optimal convergences under the balanced norm are achieved in both 1D and 2D cases. Numerical experiments support the main theoretical conclusions.
Video compression has always been a popular research area, where many traditional and deep video compression methods have been proposed. These methods typically rely on signal prediction theory to enhance compression performance by designing high efficient intra and inter prediction strategies and compressing video frames one by one. In this paper, we propose a novel model-based video compression (MVC) framework that regards scenes as the fundamental units for video sequences. Our proposed MVC directly models the intensity variation of the entire video sequence in one scene, seeking non-redundant representations instead of reducing redundancy through spatio-temporal predictions. To achieve this, we employ implicit neural representation as our basic modeling architecture. To improve the efficiency of video modeling, we first propose context-related spatial positional embedding and frequency domain supervision in spatial context enhancement. For temporal correlation capturing, we design the scene flow constrain mechanism and temporal contrastive loss. Extensive experimental results demonstrate that our method achieves up to a 20\% bitrate reduction compared to the latest video coding standard H.266 and is more efficient in decoding than existing video coding strategies.
As a popular stabilization technique, the nonsymmetric interior penalty Galerkin (NIPG) method has significant application value in computational fluid dynamics. In this paper, we study the NIPG method for a typical two-dimensional singularly perturbed convection diffusion problem on a Shishkin mesh. According to the characteristics of the solution, the mesh and numerical scheme, a new composite interpolation is introduced. In fact, this interpolation is composed of a vertices-edges-element interpolation within the layer and a local $L^{2}$-projection outside the layer. On the basis of that, by selecting penalty parameters on different types of interelement edges, we further obtain the supercloseness of almost $k+\frac{1}{2}$ order in an energy norm. Here $k$ is the degree of piecewise polynomials. Numerical tests support our theoretical conclusion.
In this paper, we study the convergence of the nonsymmetric interior penalty Galerkin (NIPG) method on a Bakhvalov-type mesh for the first time. For this purpose, a new composite interpolation is designed, which solves the inherent difficulty of analysis on Bakhvalov-type meshes. More specifically, Gauss Radau interpolation and Gauss Lobatto interpolation are used outside and inside the layer, respectively. On the basis of that, by choosing the specific values of the penalty parameters at different mesh points, we derive the supercloseness of $k+1/2$th order ($k\ge 1$), and prove the convergence of optimal order in an energy norm. The theoretical conclusion is consistent with the numerical results.
This report presents the algorithm used in the submission of Generic Event Boundary Detection (GEBD) Challenge at CVPR 2022. In this work, we improve the existing Structured Context Transformer (SC-Transformer) method for GEBD. Specifically, a transformer decoder module is added after transformer encoders to extract high quality frame features. The final classification is performed jointly on the results of the original binary classifier and a newly introduced multi-class classifier branch. To enrich motion information, optical flow is introduced as a new modality. Finally, model ensemble is used to further boost performance. The proposed method achieves 86.49% F1 score on Kinetics-GEBD test set. which improves 2.86% F1 score compared to the previous SOTA method.
In this paper, we analyze the supercloseness result of nonsymmetric interior penalty Galerkin (NIPG) method on Shishkin mesh for a singularly perturbed convection diffusion problem. According to the characteristics of the solution and the scheme, a new analysis is proposed. More specifically, Gauß Lobatto interpolation and Gauß Radau interpolation are introduced inside and outside the layer, respectively. By selecting special penalty parameters at different mesh points, we further establish supercloseness of almost k + 1 order under the energy norm. Here k is the order of piecewise polynomials. Then, a simple post processing operator is constructed. In particular, a new analysis is proposed for the stability analysis of this operator. On the basis of that, we prove that the corresponding post-processing can make the numerical solution achieve higher accuracy. Finally, superconvergence can be derived under the discrete energy norm. These theoretical conclusions can be verified numerically.
Following the performance breakthrough of denoising networks, improvements have come chiefly through novel architecture designs and increased depth. While novel denoising networks were designed for real images coming from different distributions, or for specific applications, comparatively small improvement was achieved on Gaussian denoising. The denoising solutions suffer from epistemic uncertainty that can limit further advancements. This uncertainty is traditionally mitigated through different ensemble approaches. However, such ensembles are prohibitively costly with deep networks, which are already large in size. Our work focuses on pushing the performance limits of state-of-the-art methods on Gaussian denoising. We propose a model-agnostic approach for reducing epistemic uncertainty while using only a single pretrained network. We achieve this by tapping into the epistemic uncertainty through augmented and frequency-manipulated images to obtain denoised images with varying error. We propose an ensemble method with two decoupled attention paths, over the pixel domain and over that of our different manipulations, to learn the final fusion. Our results significantly improve over the state-of-the-art baselines and across varying noise levels.
The development of neural networks has greatly improved the performance in various computer vision tasks. In the filed of image denoising, convolutional neural network based methods such as DnCNN break through the limits of classical methods, achieving better quantitative results. However, the epistemic uncertainty existing in neural networks limits further improvements in their performance over denoising tasks. Therefore, we develop and study different solutions to minimize uncertainty and further improve the removal of noise. From the perspective of ensemble learning, we implement manipulations to noisy images from the point of view of spatial and frequency domains and then denoise them using pre-trained denoising networks. We propose a fusion model consisting of two attention modules, which focus on assigning the proper weights to pixels and channels. The experimental results show that our model achieves better performance on top of the baseline of regular pre-trained denoising networks.