Searcharxiv⌕ Search

arXiv subjects

Hui Liu

Publications and source records attributed to Hui Liu.

At least 379 records · Page 21Linked to original sources
arXiv2023

Learning A Locally Unified 3D Point Cloud for View Synthesis

In this paper, we explore the problem of 3D point cloud representation-based view synthesis from a set of sparse source views. To tackle this challenging problem, we propose a new deep learning-based view synthesis paradigm that learns a locally unified 3D point cloud from source views. Specifically, we first construct sub-point clouds by projecting source views to 3D space based on their depth maps. Then, we learn the locally unified 3D point cloud by adaptively fusing points at a local neighborhood defined on the union of the sub-point clouds. Besides, we also propose a 3D geometry-guided image restoration module to fill the holes and recover high-frequency details of the rendered novel views. Experimental results on three benchmark datasets demonstrate that our method can improve the average PSNR by more than 4 dB while preserving more accurate visual details, compared with state-of-the-art view synthesis methods.

cs.CV↗
arXiv2023

A Comprehensive Survey on Trustworthy Graph Neural Networks: Privacy, Robustness, Fairness, and Explainability

Graph Neural Networks (GNNs) have made rapid developments in the recent years. Due to their great ability in modeling graph-structured data, GNNs are vastly used in various applications, including high-stakes scenarios such as financial analysis, traffic predictions, and drug discovery. Despite their great potential in benefiting humans in the real world, recent study shows that GNNs can leak private information, are vulnerable to adversarial attacks, can inherit and magnify societal bias from training data and lack interpretability, which have risk of causing unintentional harm to the users and society. For example, existing works demonstrate that attackers can fool the GNNs to give the outcome they desire with unnoticeable perturbation on training graph. GNNs trained on social networks may embed the discrimination in their decision process, strengthening the undesirable societal bias. Consequently, trustworthy GNNs in various aspects are emerging to prevent the harm from GNN models and increase the users' trust in GNNs. In this paper, we give a comprehensive survey of GNNs in the computational aspects of privacy, robustness, fairness, and explainability. For each aspect, we give the taxonomy of the related methods and formulate the general frameworks for the multiple categories of trustworthy GNNs. We also discuss the future research directions of each aspect and connections between these aspects to help achieve trustworthiness.

cs.LG↗
arXiv2023

Limit theorems for functionals of long memory linear processes with infinite variance

Let $X=\{X_n: n\in\mathbb{N}\}$ be a long memory linear process in which the coefficients are regularly varying and innovations are independent and identically distributed and belong to the domain of attraction of an $α$-stable law with $α\in (0, 2)$. Then, for any integrable and square integrable function $K$ on $\mathbb{R}$, under certain mild conditions, we establish the asymptotic behavior of the partial sum process \[ \left\{\sum\limits_{n=1}^{[Nt]}\big[K(X_n)-\E K(X_n)\big]:\; t\geq 0\right\} \] as $N$ tends to infinity, where $[Nt]$ is the integer part of $Nt$ for $t\geq 0$.

math.PR↗
arXiv2023

Generative Diffusion Models on Graphs: Methods and Applications

Diffusion models, as a novel generative paradigm, have achieved remarkable success in various image generation tasks such as image inpainting, image-to-text translation, and video generation. Graph generation is a crucial computational task on graphs with numerous real-world applications. It aims to learn the distribution of given graphs and then generate new graphs. Given the great success of diffusion models in image generation, increasing efforts have been made to leverage these techniques to advance graph generation in recent years. In this paper, we first provide a comprehensive overview of generative diffusion models on graphs, In particular, we review representative algorithms for three variants of graph diffusion models, i.e., Score Matching with Langevin Dynamics (SMLD), Denoising Diffusion Probabilistic Model (DDPM), and Score-based Generative Model (SGM). Then, we summarize the major applications of generative diffusion models on graphs with a specific focus on molecule and protein modeling. Finally, we discuss promising directions in generative diffusion models on graph-structured data. For this survey, we also created a GitHub project website by collecting the supporting resources for generative diffusion models on graphs, at the link: https://github.com/ChengyiLIU-cs/Generative-Diffusion-Models-on-Graphs

cs.LG↗
arXiv2023

Federated Learning in Big Model Era: Domain-Specific Multimodal Large Models

Multimodal data, which can comprehensively perceive and recognize the physical world, has become an essential path towards general artificial intelligence. However, multimodal large models trained on public datasets often underperform in specific industrial domains. This paper proposes a multimodal federated learning framework that enables multiple enterprises to utilize private domain data to collaboratively train large models for vertical domains, achieving intelligent services across scenarios. The authors discuss in-depth the strategic transformation of federated learning in terms of intelligence foundation and objectives in the era of big model, as well as the new challenges faced in heterogeneous data, model aggregation, performance and cost trade-off, data privacy, and incentive mechanism. The paper elaborates a case study of leading enterprises contributing multimodal data and expert knowledge to city safety operation management , including distributed deployment and efficient coordination of the federated learning platform, technical innovations on data quality improvement based on large model capabilities and efficient joint fine-tuning approaches. Preliminary experiments show that enterprises can enhance and accumulate intelligent capabilities through multimodal model federated learning, thereby jointly creating an smart city model that provides high-quality intelligent services covering energy infrastructure safety, residential community security, and urban operation management. The established federated learning cooperation ecosystem is expected to further aggregate industry, academia, and research resources, realize large models in multiple vertical domains, and promote the large-scale industrial application of artificial intelligence and cutting-edge research on multimodal federated learning.

cs.LG↗
arXiv2023

ICL-D3IE: In-Context Learning with Diverse Demonstrations Updating for Document Information Extraction

Large language models (LLMs), such as GPT-3 and ChatGPT, have demonstrated remarkable results in various natural language processing (NLP) tasks with in-context learning, which involves inference based on a few demonstration examples. Despite their successes in NLP tasks, no investigation has been conducted to assess the ability of LLMs to perform document information extraction (DIE) using in-context learning. Applying LLMs to DIE poses two challenges: the modality and task gap. To this end, we propose a simple but effective in-context learning framework called ICL-D3IE, which enables LLMs to perform DIE with different types of demonstration examples. Specifically, we extract the most difficult and distinct segments from hard training documents as hard demonstrations for benefiting all test instances. We design demonstrations describing relationships that enable LLMs to understand positional relationships. We introduce formatting demonstrations for easy answer extraction. Additionally, the framework improves diverse demonstrations by updating them iteratively. Our experiments on three widely used benchmark datasets demonstrate that the ICL-D3IE framework enables Davinci-003/ChatGPT to achieve superior performance when compared to previous pre-trained methods fine-tuned with full training in both the in-distribution (ID) setting and in the out-of-distribution (OOD) setting. Code is available at https://github.com/MAEHCM/ICL-D3IE.

cs.CL↗
arXiv2023

Pion quasiparticles in isospin medium from holography

The properties of the pion quasiparticle in hot and dense isospin medium, including the temperature and isospin chemical potential dependence of their screening mass, pole mass and thermal width, as well as their relationships with the pion superfluid phase transition, are investigated in the framework of two-flavor ($N_{f}=2$) soft-wall AdS/QCD models. We extract the screening mass of the pion from the pole of the spatial two-point Retarded correlation function. We find that the screening masses of both neutral and charged pions increase monotonously with the increasing of temperature. However, the isospin chemical potential $μ_{I}$ would depress the screening masses of the charged pions, $m_{π^{\pm},\rm{scr}}$. With the increasing of $μ_{I}$, $m_{π^{\pm},\rm{scr}}$ monotonically decrease to zero on the boundary between the normal phase and the pion superfluid phase, while the screening mass of the neutral pion, $m_{π^0,\rm{scr}}$, remains almost unchanged. The pole mass $m_{\rm{pole}}$ and thermal width $Γ$ of the pion are extracted from the pole of temporal two-point Retarded correlation function, i.e., the corresponding quasi-normal frequencies, $ω=m_{\rm{pole}}-iΓ/2$. The results show that the pole masses of the three modes ($π^0, π^+, π^-$) are splitting at finite $μ_{I}$. The thermal widths of the three modes monotonically increase with temperature. Furthermore, the pole mass of $π^+$ decreases almost linearly with the increasing of $μ_{I}$ and reaches zero at $μ_{I}=μ_{I}^c$, It means that $π^+$ becomes a massless Goldstone boson of the pion superfluid phase transition.

hep-ph↗
arXiv2023

Improving Text Semantic Similarity Modeling through a 3D Siamese Network

Siamese networks have gained popularity as a method for modeling text semantic similarity. Traditional methods rely on pooling operation to compress the semantic representations from Transformer blocks in encoding, resulting in two-dimensional semantic vectors and the loss of hierarchical semantic information from Transformer blocks. Moreover, this limited structure of semantic vectors is akin to a flattened landscape, which restricts the methods that can be applied in downstream modeling, as they can only navigate this flat terrain. To address this issue, we propose a novel 3D Siamese network for text semantic similarity modeling, which maps semantic information to a higher-dimensional space. The three-dimensional semantic tensors not only retains more precise spatial and feature domain information but also provides the necessary structural condition for comprehensive downstream modeling strategies to capture them. Leveraging this structural advantage, we introduce several modules to reinforce this 3D framework, focusing on three aspects: feature extraction, attention, and feature fusion. Our extensive experiments on four text semantic similarity benchmarks demonstrate the effectiveness and efficiency of our 3D Siamese Network.

cs.CL↗
arXiv2023

Large deviations of invariant measure for the 3D stochastic hyperdissipative Navier-Stokes equations

In this paper, we consider the large deviations of invariant measure for the 3D stochastic hyperdissipative Navier-Stokes equations driven by additive noise. The unique ergodicity of invariant measure as a preliminary result is proved using a deterministic argument by the exponential moment and exponential stability estimates. Then, the uniform large deviations is established by the uniform contraction principle. Finally, using the unique ergodicity and the uniform large deviations results, we prove the large deviations of invariant measure by verifying the Freidlin-Wentzell large deviations upper and lower bounds.

math.AP↗
arXiv2023

Mixed higher-order topology: boundary non-Hermitian skin effect induced by a Floquet bulk

We show that anomalous Floquet topological insulators generate intrinsic, non-Hermitian topology on their boundary. As a consequence, removing a boundary hopping from the time-evolution operator stops the propagation of chiral edge modes, leading to a non-Hermitian skin effect. This does not occur in Floquet Chern insulators, however, in which boundary modes continue propagating. The non-Hermitian skin effect on the boundary is a consequence of the nontrivial topology of the bulk Floquet operator, which we show by designing a real-space topological invariant. Our work introduces a form of `mixed' higher-order topology, providing a bridge between research on periodically-driven systems and the study of non-Hermiticity. It suggests that periodic driving, which has already been demonstrated in a wide range of experiments, may be used to generate non-Hermitian skin effects.

cond-mat.mes-hall↗
arXiv2023

Light Field Reconstruction via Deep Adaptive Fusion of Hybrid Lenses

This paper explores the problem of reconstructing high-resolution light field (LF) images from hybrid lenses, including a high-resolution camera surrounded by multiple low-resolution cameras. The performance of existing methods is still limited, as they produce either blurry results on plain textured areas or distortions around depth discontinuous boundaries. To tackle this challenge, we propose a novel end-to-end learning-based approach, which can comprehensively utilize the specific characteristics of the input from two complementary and parallel perspectives. Specifically, one module regresses a spatially consistent intermediate estimation by learning a deep multidimensional and cross-domain feature representation, while the other module warps another intermediate estimation, which maintains the high-frequency textures, by propagating the information of the high-resolution view. We finally leverage the advantages of the two intermediate estimations adaptively via the learned attention maps, leading to the final high-resolution LF image with satisfactory results on both plain textured areas and depth discontinuous boundaries. Besides, to promote the effectiveness of our method trained with simulated hybrid data on real hybrid data captured by a hybrid LF imaging system, we carefully design the network architecture and the training strategy. Extensive experiments on both real and simulated hybrid data demonstrate the significant superiority of our approach over state-of-the-art ones. To the best of our knowledge, this is the first end-to-end deep learning method for LF reconstruction from a real hybrid input. We believe our framework could potentially decrease the cost of high-resolution LF data acquisition and benefit LF data storage and transmission.

eess.IV↗
arXiv2023

Integrated Broadband Mode Division Demultiplexer in Waveguide Arrays

On-chip mode division (de)multiplexing plays a significant role in the integrated devices to greatly improve communication capacity. One of the pursuing goals of mode division (de)multiplexer is how to realize such devices with larger bandwidth and better fabrication tolerance. However, integrated broadband mode division (de)multiplexing devices were rarely reported up to now. In this work, we experimentally realize a broadband mode division demultiplexing device with a high degree of fabrication tolerance based on lithium niobate-on-insulator waveguide array. By taking advantage of the fact that different modes in the waveguide array have different group velocity, we experimentally confirmed that different modes in the waveguide array have different transverse displacements. The implementation of the integrated broadband mode division demultiplexer may have future applications in information processing technology, quantum communication as well as quantum computing.

physics.optics↗
arXiv2023

Do-GOOD: Towards Distribution Shift Evaluation for Pre-Trained Visual Document Understanding Models

Numerous pre-training techniques for visual document understanding (VDU) have recently shown substantial improvements in performance across a wide range of document tasks. However, these pre-trained VDU models cannot guarantee continued success when the distribution of test data differs from the distribution of training data. In this paper, to investigate how robust existing pre-trained VDU models are to various distribution shifts, we first develop an out-of-distribution (OOD) benchmark termed Do-GOOD for the fine-Grained analysis on Document image-related tasks specifically. The Do-GOOD benchmark defines the underlying mechanisms that result in different distribution shifts and contains 9 OOD datasets covering 3 VDU related tasks, e.g., document information extraction, classification and question answering. We then evaluate the robustness and perform a fine-grained analysis of 5 latest VDU pre-trained models and 2 typical OOD generalization algorithms on these OOD datasets. Results from the experiments demonstrate that there is a significant performance gap between the in-distribution (ID) and OOD settings for document images, and that fine-grained analysis of distribution shifts can reveal the brittle nature of existing pre-trained VDU models and OOD generalization algorithms. The code and datasets for our Do-GOOD benchmark can be found at https://github.com/MAEHCM/Do-GOOD.

cs.CV↗
arXiv2023

Singularities of divisors of small degrees on abelian varieties

Pareschi showed that an ample divisor of degree $d$ on a simple abelian variety of dimension $g$ has mild singularities when $d<g$. We extend his result to general polarized abelian varieties. We also show that ample divisors of degree $3$ or $4$ on an abelian variety have mild singularities.

math.AG↗
arXiv2023

Self-Explainable Graph Neural Networks for Link Prediction

Graph Neural Networks (GNNs) have achieved state-of-the-art performance for link prediction. However, GNNs suffer from poor interpretability, which limits their adoptions in critical scenarios that require knowing why certain links are predicted. Despite various methods proposed for the explainability of GNNs, most of them are post-hoc explainers developed for explaining node classification. Directly adopting existing post-hoc explainers for explaining link prediction is sub-optimal because: (i) post-hoc explainers usually adopt another strategy or model to explain a target model, which could misinterpret the target model; and (ii) GNN explainers for node classification identify crucial subgraphs around each node for the explanation; while for link prediction, one needs to explain the prediction for each pair of nodes based on graph structure and node attributes. Therefore, in this paper, we study a novel problem of self-explainable GNNs for link prediction, which can simultaneously give accurate predictions and explanations. Concretely, we propose a new framework and it can find various $K$ important neighbors of one node to learn pair-specific representations for links from this node to other nodes. These $K$ different neighbors represent important characteristics of the node and model various factors for links from it. Thus, $K$ neighbors can provide explanations for the existence of links. Experiments on both synthetic and real-world datasets verify the effectiveness of the proposed framework for link prediction and explanation.

cs.LG↗
arXiv2023

Multiple symmetry protected BIC lines in two dimensional synthetic parameter space

Bound states in the continuum (BICs) have attracted significant interest in recent years due to their unique optical properties, such as infinite quality factor and wave localization. In order to improve the optical performance of BICs based devices, more degrees of freedom are required to tune BICs in high-dimension parameter space for practical applications. To effectively tune more BICs, we form a 2D synthetic parameter space based on a nanohole metasurface array. Multiple symmetry protected BIC modes with high Q factors can be achieved at high-order symmetry point. Through manipulating asymmetry parameters, BIC lines formed by a series of BIC modes can be found in the 2D synthetic parameter space. Moreover, the electric field distributions are investigated to demonstrate the generation and evolution of BICs. By measuring the absorption spectra, the tuning of multiple BICs with synthet-ic asymmetry parameters is experimentally explored, which agrees well with theoretical results. Therefore, our de-sign can provide new insight for a variety of on-chip applications, such as non-linear devices, integrated nanolasing array and high-resolution sensors for infrared molecular detection.

physics.optics↗
arXiv2023

TCR: Short Video Title Generation and Cover Selection with Attention Refinement

With the widespread popularity of user-generated short videos, it becomes increasingly challenging for content creators to promote their content to potential viewers. Automatically generating appealing titles and covers for short videos can help grab viewers' attention. Existing studies on video captioning mostly focus on generating factual descriptions of actions, which do not conform to video titles intended for catching viewer attention. Furthermore, research for cover selection based on multimodal information is sparse. These problems motivate the need for tailored methods to specifically support the joint task of short video title generation and cover selection (TG-CS) as well as the demand for creating corresponding datasets to support the studies. In this paper, we first collect and present a real-world dataset named Short Video Title Generation (SVTG) that contains videos with appealing titles and covers. We then propose a Title generation and Cover selection with attention Refinement (TCR) method for TG-CS. The refinement procedure progressively selects high-quality samples and highly relevant frames and text tokens within each sample to refine model training. Extensive experiments show that our TCR method is superior to various existing video captioning methods in generating titles and is able to select better covers for noisy real-world short videos.

cs.CV↗
arXiv2023

Thermalization and prethermalization in the soft-wall AdS/QCD model

The real-time dynamics of chiral phase transition is investigated in a two-flavor ($N_f=2$) soft-wall AdS/QCD model. To understand the dynamics of thermalization, we quench the system from initial states deviating from the equilibrium states. Then, we solve the nonequilibrium evolution of the order parameter (chiral condensate $\langle σ\equiv\bar{q}q\rangle$). It is shown that the system undergoes an exponential relaxation at temperatures away from the critical temperature $T_c$. The relaxation time diverges at $T_c$, presenting a typical behavior of critical slowing down. Numerically, we extract the dynamic critical exponent $z$, and get $z\approx 2$ by fitting the scaling behavior $σ\propto t^{-β/(νz)}$, where the mean-field static critical exponents (order parameter critical exponent $β=1/2$, correlation length critical exponent $ν=1/2$ ) have been applied. More interestingly, it is remarked that, for a large class of initial states, the system would linger over a quasi-steady state for a certain period of time before the thermalization. It is suggested that the interesting phenomenon, known as prethermalization, has been observed in the framework of holographic models. In such prethermal stage, we verify that the system is characterized by a universal dynamical scaling law and described by the initial-slip exponent $θ=0$.

hep-ph↗