arXiv subjects

Jie Wu

Publications and source records attributed to Jie Wu.

At least 199 records · Page 11Linked to original sources

arXiv2023

Knot data analysis using multiscale Gauss link integral

In the past decade, topological data analysis (TDA) has emerged as a powerful approach in data science. The main technique in TDA is persistent homology, which tracks topological invariants over the filtration of point cloud data using algebraic topology. Although knot theory and related subjects are a focus of study in mathematics, their success in practical applications is quite limited due to the lack of localization and quantization. We address these challenges by introducing knot data analysis (KDA), a new paradigm that incorporating curve segmentation and multiscale analysis into the Gauss link integral. The resulting multiscale Gauss link integral (mGLI) recovers the global topological properties of knots and links at an appropriate scale but offers multiscale feature vectors to capture the local structures and connectivities of each curve segment at various scales. The proposed mGLI significantly outperforms other state-of-the-art methods in benchmark protein flexibility analysis, including earlier persistent homology-based methods. Our approach enables the integration of artificial intelligence (AI) and KDA for general curve-like objects and data.

math.GT↗

arXiv2023

AutoDiffusion: Training-Free Optimization of Time Steps and Architectures for Automated Diffusion Model Acceleration

Diffusion models are emerging expressive generative models, in which a large number of time steps (inference steps) are required for a single image generation. To accelerate such tedious process, reducing steps uniformly is considered as an undisputed principle of diffusion models. We consider that such a uniform assumption is not the optimal solution in practice; i.e., we can find different optimal time steps for different models. Therefore, we propose to search the optimal time steps sequence and compressed model architecture in a unified framework to achieve effective image generation for diffusion models without any further training. Specifically, we first design a unified search space that consists of all possible time steps and various architectures. Then, a two stage evolutionary algorithm is introduced to find the optimal solution in the designed search space. To further accelerate the search process, we employ FID score between generated and real samples to estimate the performance of the sampled examples. As a result, the proposed method is (i).training-free, obtaining the optimal time steps and model architecture without any training process; (ii). orthogonal to most advanced diffusion samplers and can be integrated to gain better sample quality. (iii). generalized, where the searched time steps and architectures can be directly applied on different diffusion models with the same guidance scale. Experimental results show that our method achieves excellent performance by using only a few time steps, e.g. 17.86 FID score on ImageNet 64 $\times$ 64 with only four steps, compared to 138.66 with DDIM. The code is available at https://github.com/lilijiangg/AutoDiffusion.

cs.CV↗

arXiv2023

UGC: Unified GAN Compression for Efficient Image-to-Image Translation

Recent years have witnessed the prevailing progress of Generative Adversarial Networks (GANs) in image-to-image translation. However, the success of these GAN models hinges on ponderous computational costs and labor-expensive training data. Current efficient GAN learning techniques often fall into two orthogonal aspects: i) model slimming via reduced calculation costs; ii)data/label-efficient learning with fewer training data/labels. To combine the best of both worlds, we propose a new learning paradigm, Unified GAN Compression (UGC), with a unified optimization objective to seamlessly prompt the synergy of model-efficient and label-efficient learning. UGC sets up semi-supervised-driven network architecture search and adaptive online semi-supervised distillation stages sequentially, which formulates a heterogeneous mutual learning scheme to obtain an architecture-flexible, label-efficient, and performance-excellent model.

cs.CV↗

arXiv2023

DiffusionEngine: Diffusion Model is Scalable Data Engine for Object Detection

Data is the cornerstone of deep learning. This paper reveals that the recently developed Diffusion Model is a scalable data engine for object detection. Existing methods for scaling up detection-oriented data often require manual collection or generative models to obtain target images, followed by data augmentation and labeling to produce training pairs, which are costly, complex, or lacking diversity. To address these issues, we presentDiffusionEngine (DE), a data scaling-up engine that provides high-quality detection-oriented training pairs in a single stage. DE consists of a pre-trained diffusion model and an effective Detection-Adapter, contributing to generating scalable, diverse and generalizable detection data in a plug-and-play manner. Detection-Adapter is learned to align the implicit semantic and location knowledge in off-the-shelf diffusion models with detection-aware signals to make better bounding-box predictions. Additionally, we contribute two datasets, i.e., COCO-DE and VOC-DE, to scale up existing detection benchmarks for facilitating follow-up research. Extensive experiments demonstrate that data scaling-up via DE can achieve significant improvements in diverse scenarios, such as various detection algorithms, self-supervised pre-training, data-sparse, label-scarce, cross-domain, and semi-supervised learning. For example, when using DE with a DINO-based adapter to scale up data, mAP is improved by 3.1% on COCO, 7.6% on VOC, and 11.5% on Clipart.

cs.CV↗

arXiv2023

DLIP: Distilling Language-Image Pre-training

Vision-Language Pre-training (VLP) shows remarkable progress with the assistance of extremely heavy parameters, which challenges deployment in real applications. Knowledge distillation is well recognized as the essential procedure in model compression. However, existing knowledge distillation techniques lack an in-depth investigation and analysis of VLP, and practical guidelines for VLP-oriented distillation are still not yet explored. In this paper, we present DLIP, a simple yet efficient Distilling Language-Image Pre-training framework, through which we investigate how to distill a light VLP model. Specifically, we dissect the model distillation from multiple dimensions, such as the architecture characteristics of different modules and the information transfer of different modalities. We conduct comprehensive experiments and provide insights on distilling a light but performant VLP model. Experimental results reveal that DLIP can achieve a state-of-the-art accuracy/efficiency trade-off across diverse cross-modal tasks, e.g., image-text retrieval, image captioning and visual question answering. For example, DLIP compresses BLIP by 1.9x, from 213M to 108M parameters, while achieving comparable or better performance. Furthermore, DLIP succeeds in retaining more than 95% of the performance with 22.4% parameters and 24.8% FLOPs compared to the teacher model and accelerates inference speed by 2.7x.

cs.CV↗

arXiv2023

AlignDet: Aligning Pre-training and Fine-tuning in Object Detection

The paradigm of large-scale pre-training followed by downstream fine-tuning has been widely employed in various object detection algorithms. In this paper, we reveal discrepancies in data, model, and task between the pre-training and fine-tuning procedure in existing practices, which implicitly limit the detector's performance, generalization ability, and convergence speed. To this end, we propose AlignDet, a unified pre-training framework that can be adapted to various existing detectors to alleviate the discrepancies. AlignDet decouples the pre-training process into two stages, i.e., image-domain and box-domain pre-training. The image-domain pre-training optimizes the detection backbone to capture holistic visual abstraction, and box-domain pre-training learns instance-level semantics and task-aware concepts to initialize the parts out of the backbone. By incorporating the self-supervised pre-trained backbones, we can pre-train all modules for various detectors in an unsupervised paradigm. As depicted in Figure 1, extensive experiments demonstrate that AlignDet can achieve significant improvements across diverse protocols, such as detection algorithm, model backbone, data setting, and training schedule. For example, AlignDet improves FCOS by 5.3 mAP, RetinaNet by 2.1 mAP, Faster R-CNN by 3.3 mAP, and DETR by 2.3 mAP under fewer epochs.

cs.CV↗

arXiv2023

The magnitude homology of a hypergraph

The magnitude homology, introduced by R. Hepworth and S. Willerton, offers a topological invariant that enables the study of graph properties. Hypergraphs, being a generalization of graphs, serve as popular mathematical models for data with higher-order structures. In this paper, we focus on describing the topological characteristics of hypergraphs by considering their magnitude homology. We begin by examining the distances between hyperedges in a hypergraph and establish the magnitude homology of hypergraphs. Additionally, we explore the relationship between the magnitude and the magnitude homology of hypergraphs. Furthermore, we derive several functorial properties of the magnitude homology for hypergraphs. Lastly, we present the Künneth theorem for the simple magnitude homology of hypergraphs.

math.AT↗

arXiv2023

SepViT: Separable Vision Transformer

Vision Transformers have witnessed prevailing success in a series of vision tasks. However, these Transformers often rely on extensive computational costs to achieve high performance, which is burdensome to deploy on resource-constrained devices. To alleviate this issue, we draw lessons from depthwise separable convolution and imitate its ideology to design an efficient Transformer backbone, i.e., Separable Vision Transformer, abbreviated as SepViT. SepViT helps to carry out the local-global information interaction within and among the windows in sequential order via a depthwise separable self-attention. The novel window token embedding and grouped self-attention are employed to compute the attention relationship among windows with negligible cost and establish long-range visual interactions across multiple windows, respectively. Extensive experiments on general-purpose vision benchmarks demonstrate that SepViT can achieve a state-of-the-art trade-off between performance and latency. Among them, SepViT achieves 84.2% top-1 accuracy on ImageNet-1K classification while decreasing the latency by 40%, compared to the ones with similar accuracy (e.g., CSWin). Furthermore, SepViT achieves 51.0% mIoU on ADE20K semantic segmentation task, 47.9 AP on the RetinaNet-based COCO detection task, 49.4 box AP and 44.6 mask AP on Mask R-CNN-based COCO object detection and instance segmentation tasks.

cs.CV↗

arXiv2023

A sufficient condition for the existence of fractional $(g,f,n)$-critical covered graphs

In data transmission networks, the availability of data transmission is equivalent to the existence of the fractional factor of the corresponding graph which is generated by the network. Research on the existence of fractional factors under specific network structures can help scientists design and construct networks with high data transmission rates. A graph $G$ is called a fractional $(g,f)$-covered graph if for any $e\in E(G)$, $G$ admits a fractional $(g,f)$-factor covering $e$. A graph $G$ is called a fractional $(g,f,n)$-critical covered graph if after removing any $n$ vertices of $G$, the resulting graph of $G$ is a fractional $(g,f)$-covered graph. In this paper, we verify that if a graph $G$ of order $p$ satisfies $p\geq\frac{(a+b-1)(a+b-2)+(a+d)n+1}{a+d}$, $δ(G)\geq\frac{(b-d-1)p+(a+d)n+a+b+1}{a+b-1}$ and $δ(G)>\frac{(b-d-2)p+2α(G)+(a+d)n+1}{a+b-2}$, then $G$ is a fractional $(g,f,n)$-critical covered graph, where $g,f:V(G)\rightarrow Z^{+}$ be two functions such that $a\leq g(x)\leq f(x)-d\leq b-d$ for all $x\in V(G)$, which is a generalization of Zhou's previous result [S. Zhou, Some new sufficient conditions for graphs to have fractional $k$-factors, International Journal of Computer Mathematics 88(3)(2011)484--490].

math.CO↗

arXiv2023

The algebraic stability for persistent Laplacians

The stability of topological persistence is one of the fundamental issues in topological data analysis. Numerous methods have been proposed to address the stability of persistent modules or persistence diagrams. Recently, the concept of persistent Laplacians has emerged as a novel approach to topological persistence, attracting significant attention and finding applications in various fields. In this paper, we investigate the stability of persistent Laplacians. We introduce the notion of ``Laplacian trees'', which captures the collection of persistent Laplacians that persist from a given parameter. To formalize our study, we construct the category of Laplacian trees and establish an algebraic stability theorem for persistent Laplacian trees. Notably, our stability theorem is applied to the real-valued functions on simplicial complexes and digraphs.

math.AT↗

arXiv2023

When Decentralized Optimization Meets Federated Learning

Federated learning is a new learning paradigm for extracting knowledge from distributed data. Due to its favorable properties in preserving privacy and saving communication costs, it has been extensively studied and widely applied to numerous data analysis applications. However, most existing federated learning approaches concentrate on the centralized setting, which is vulnerable to a single-point failure. An alternative strategy for addressing this issue is the decentralized communication topology. In this article, we systematically investigate the challenges and opportunities when renovating decentralized optimization for federated learning. In particular, we discussed them from the model, data, and communication sides, respectively, which can deepen our understanding about decentralized federated learning.

cs.LG↗

arXiv2023

Weighted (Co)homology and Weighted Laplacian

In this paper, we generalize the combinatorial Laplace operator of Horak and Jost by introducing the $ϕ$-weighted coboundary operator induced by a weight function $ϕ$. Our weight function $ϕ$ is a generalization of Dawson's weighted boundary map. We show that our above-mentioned generalizations include new cases that are not covered by previous literature. Our definition of weighted Laplacian for weighted simplicial complexes is also applicable to weighted/unweighted graphs and digraphs.

math.AT↗

arXiv2023

Persistent hyperdigraph homology and persistent hyperdigraph Laplacians

Hypergraphs are useful mathematical models for describing complex relationships among members of a structured graph, while hyperdigraphs serve as a generalization that can encode asymmetric relationships in the data. However, obtaining topological information directly from hyperdigraphs remains a challenge. To address this issue, we introduce hyperdigraph homology in this work. We also propose topological hyperdigraph Laplacians, which can extract both harmonic spectra and non-harmonic spectra from directed and internally organized data. Moreover, we introduce persistent hyperdigraph homology and persistent hyperdigraph Laplacians through filtration, enabling the capture of topological persistence and homotopic shape evolution of directed and structured data across multiple scales. The proposed methods offer new multiscale algebraic topology tools for topological data analysis.

math.AT↗

arXiv2023

FreeSeg: Unified, Universal and Open-Vocabulary Image Segmentation

Recently, open-vocabulary learning has emerged to accomplish segmentation for arbitrary categories of text-based descriptions, which popularizes the segmentation system to more general-purpose application scenarios. However, existing methods devote to designing specialized architectures or parameters for specific segmentation tasks. These customized design paradigms lead to fragmentation between various segmentation tasks, thus hindering the uniformity of segmentation models. Hence in this paper, we propose FreeSeg, a generic framework to accomplish Unified, Universal and Open-Vocabulary Image Segmentation. FreeSeg optimizes an all-in-one network via one-shot training and employs the same architecture and parameters to handle diverse segmentation tasks seamlessly in the inference procedure. Additionally, adaptive prompt learning facilitates the unified model to capture task-aware and category-sensitive concepts, improving model robustness in multi-task and varied scenarios. Extensive experimental results demonstrate that FreeSeg establishes new state-of-the-art results in performance and generalization on three segmentation tasks, which outperforms the best task-specific architectures by a large margin: 5.5% mIoU on semantic segmentation, 17.6% mAP on instance segmentation, 20.1% PQ on panoptic segmentation for the unseen class on COCO.

cs.CV↗

arXiv2023

A Comprehensive Study on Optimizing Systems with Data Processing Units

New hardware, such as SmartNICs, has been released to offload network applications in data centers. Off-path SmartNICs, a type of multi-core SoC SmartNICs, have attracted the attention of many researchers. Unfortunatelly, they lack the fully exploration of off-path SmartNICs. In this paper, we use a BlueField SmartNIC as an example to conduct a systematical study on the advantages and disadvantages of off-path SmartNICs. We make a detailed performance characterization on an off-path SmartNIC including computing power and network communication overhead, and propose the following advices: 1) Directly utilize the specific accelerators on the SmartNIC to offload applications; 2) Offload latency-insensitive background processing to the SmartNIC to reduce the load on the host; 3) Regard the SmartNIC as a new endpoint in the network to expand the computing power and storage resources of the server host; 4) Avoid directly employing the design method for systems based on on-path SmartNICs. We apply these advices to several use cases and show the performance improvements.

cs.NI↗

arXiv2023

The stability of persistent homology of hypergraphs

Hypergraph is the most general model for complex networks involving group interactions. Taking the ideas of path homology from Alexander Grigor'yan, Yong Lin, Yuri Muranov and Shing-Tung Yau [18-22], Stephane Bressan, Jingyan Li and the authors of this article introduced embedded homology of hypergraphs [6] in 2019, which has leaded to successful applications in protein-ligand binding network [24, 25] in 2021. A fundamental question arising from practical applications is about the stability of the persistent embedded homology of hypergraphs. In this paper, we prove the stability of the persistent embedded homology as well as the persistent homology of the associated simplicial complex with respect to perturbations of the filtration on a hypergraph. We apply the persistent homology methods to morphisms of hypergraphs and prove the stability with respect to perturbations of the filtrations. We prove the constancy of the persistent Betti numbers under some conditions on the simple-homotopy types of hypergraphs.

math.AT↗

arXiv2022

Masked Vision-Language Transformers for Scene Text Recognition

Scene text recognition (STR) enables computers to recognize and read the text in various real-world scenes. Recent STR models benefit from taking linguistic information in addition to visual cues into consideration. We propose a novel Masked Vision-Language Transformers (MVLT) to capture both the explicit and the implicit linguistic information. Our encoder is a Vision Transformer, and our decoder is a multi-modal Transformer. MVLT is trained in two stages: in the first stage, we design a STR-tailored pretraining method based on a masking strategy; in the second stage, we fine-tune our model and adopt an iterative correction method to improve the performance. MVLT attains superior results compared to state-of-the-art STR models on several benchmarks. Our code and model are available at https://github.com/onealwj/MVLT.

cs.CV↗

arXiv2022

Towards the assignments for $1^{1}D_{2}$ and $1^{3}D_{2}$ meson nonets

In this work, we investigate the mass spectrum of $1^{1}D_{2}$ and $1^{3}D_{2}$ meson nonets in the framework of the meson mass matrix and Regge phenomenology. The results are compared with the values from different phenomenological models and may be useful for the assignment of the $1^{1}D_{2}$ and $1^{3}D_{2}$ meson nonets in the future.

hep-ph↗