arXiv subjects

Hanqing Zhao

Publications and source records attributed to Hanqing Zhao.

At least 37 records · Page 2Linked to original sources

arXiv2024

"Hall" transport of liquid crystal solitons in Couette flow

Topology establishes a unifying framework for a diverse range of scientific areas including particle physics, cosmology, and condensed matter physics. One of the most fascinating manifestations of topology in the context of condensed matter is the topological Hall effect, and its relative: the Skyrmion Hall effect. Skyrmions are stable vortex-like spin configurations in certain chiral magnets, and when subject to external electric currents can drift in the transverse direction to the current. These quasi-particles are characterised by a conserved topological charge which in the Skyrmion Hall effect plays the role of electric charges in the ordinary Hall effect. Recently, it has been shown that liquid crystals endowed with chiral properties serve as an ideal testbed for the fundamental investigation of topological solitons, including their two- and three-dimensional realisations. Here, we show experimentally and numerically that three-dimensional solitons aka "torons" exhibit a Hall-like effect when driven by shear flows: the torons are deflected in the direction perpendicular to the shear plane. The experimental results are rationalised in terms of the dynamic Ericksen-Leslie equations, which predict the emergence of the transverse component of the net mass flow, the magnitude of which scales as the 3rd power of the shear rate. The perturbation analysis highlights an interplay of the viscous and chiral elastic torques as the mechanism for the emergence of net transverse currents. Numerical simulations demonstrate, however, that torons are not merely dragged by the flow but move with their own transverse speed, much larger than the average flow velocity in the transverse direction. Our findings may enable responsive microfluidic applications relying on soft topological solitons.

cond-mat.soft↗

arXiv2024

Construction of a Syntactic Analysis Map for Yi Shui School through Text Mining and Natural Language Processing Research

Entity and relationship extraction is a crucial component in natural language processing tasks such as knowledge graph construction, question answering system design, and semantic analysis. Most of the information of the Yishui school of traditional Chinese Medicine (TCM) is stored in the form of unstructured classical Chinese text. The key information extraction of TCM texts plays an important role in mining and studying the academic schools of TCM. In order to solve these problems efficiently using artificial intelligence methods, this study constructs a word segmentation and entity relationship extraction model based on conditional random fields under the framework of natural language processing technology to identify and extract the entity relationship of traditional Chinese medicine texts, and uses the common weighting technology of TF-IDF information retrieval and data mining to extract important key entity information in different ancient books. The dependency syntactic parser based on neural network is used to analyze the grammatical relationship between entities in each ancient book article, and it is represented as a tree structure visualization, which lays the foundation for the next construction of the knowledge graph of Yishui school and the use of artificial intelligence methods to carry out the research of TCM academic schools.

cs.CL↗

arXiv2023

Topological solitonic macromolecules

Being ubiquitous, solitons have particle-like properties, exhibiting behaviour often associated with atoms. Bound solitons emulate dynamics of molecules, though solitonic analogues of polymeric materials have not been considered yet. Here we experimentally create and model soliton polymers, which we call polyskyrmionomers, built of atom-like individual solitons characterized by the topological invariant representing the skyrmion number. With the help of nonlinear optical imaging and numerical modelling based on minimizing the free energy, we reveal how topological point defects bind the solitonic quasi-atoms into polyskyrmionomers, featuring linear, branched, and other macromolecule-resembling architectures, as well as allowing for encoding data by spatial distributions of the skyrmion number. Application of oscillating electric fields activates diverse modes of locomotion and internal vibrations of these self-assembled soliton structures, which depend on symmetry of the solitonic macromolecules. Our findings suggest new designs of soliton meta matter, with a potential for the use in fundamental research and technology.

nlin.PS↗

arXiv2023

Sharp estimates and non-degeneracy of low energy nodal solutions for the Lane-Emden equation in dimension two

We study the Lane-Emden problem \[\begin{cases} -Δu_p =|u_p|^{p-1}u_p&\text{in}\quad Ω, u_p=0 &\text{on}\quad\partialΩ, \end{cases}\] where $Ω\subset\mathbb R^2$ is a smooth bounded domain and $p>1$ is sufficiently large. We obtain sharp estimates and non-degeneracy of low energy nodal solutions $u_p$ (i.e. nodal solutions satisfying $\lim_{p\to+\infty}p\int_Ω|u_p|^{p+1}dx=16πe$). As applications, we prove that the comparable condition $p(\|u_p^+\|_{\infty}-\|u_p^-\|_{\infty})=O(1)$ holds automatically for least energy nodal solutions, which confirms a conjecture raised by (Grossi-Grumiau-Pacella, Ann.I.H. Poincare-AN, 30 (2013), 121-140) and (Grossi-Grumiau-Pacella, J.Math.Pures Appl. 101 (2014), 735-754).

math.AP↗

arXiv2023

Asymptotic behavior of least energy nodal solutions for biharmonic Lane-Emden problems in dimension four

In this paper, we study the asymptotic behavior of least energy nodal solutions $u_p(x)$ to the following fourth-order elliptic problem \[ \begin{cases} Δ^2 u =|u|^{p-1}u \quad &\hbox{in}\;Ω, \\ u=\frac{\partial u}{\partial ν}=0 \ \ &\hbox{on}\;\partialΩ, \end{cases} \] where $Ω$ is a bounded $C^{4,α}$ domain in $\mathbb{R}^4$ and $p>1$. Among other things, we show that up to a subsequence of $p\to+\infty$, $pu_p(x)\to 64π^2\sqrt{e}(G(x,x^+)-G(x,x^-))$, where $x^+\neq x^-\in Ω$ and $G(x,y)$ is the corresponding Green function of $Δ^2$. This generalize those results for $-Δu=|u|^{p-1}u$ in dimension two by (Grossi-Grumiau-Pacella, Ann.I.H.Poincaré-AN, 30 (2013), 121-140) to the biharmonic case, and also gives an alternative proof of Grossi-Grumiau-Pacella's results without assuming their comparable condition $p(\|u_p^+\|_{\infty}-\|u_p^-\|_{\infty})=O(1)$.

math.AP↗

arXiv2023

X-Paste: Revisiting Scalable Copy-Paste for Instance Segmentation using CLIP and StableDiffusion

Copy-Paste is a simple and effective data augmentation strategy for instance segmentation. By randomly pasting object instances onto new background images, it creates new training data for free and significantly boosts the segmentation performance, especially for rare object categories. Although diverse, high-quality object instances used in Copy-Paste result in more performance gain, previous works utilize object instances either from human-annotated instance segmentation datasets or rendered from 3D object models, and both approaches are too expensive to scale up to obtain good diversity. In this paper, we revisit Copy-Paste at scale with the power of newly emerged zero-shot recognition models (e.g., CLIP) and text2image models (e.g., StableDiffusion). We demonstrate for the first time that using a text2image model to generate images or zero-shot recognition model to filter noisily crawled images for different object categories is a feasible way to make Copy-Paste truly scalable. To make such success happen, we design a data acquisition and processing framework, dubbed ``X-Paste", upon which a systematic study is conducted. On the LVIS dataset, X-Paste provides impressive improvements over the strong baseline CenterNet2 with Swin-L as the backbone. Specifically, it archives +2.6 box AP and +2.1 mask AP gains on all classes and even more significant gains with +6.8 box AP, +6.5 mask AP on long-tail classes. Our code and models are available at https://github.com/yoctta/XPaste.

cs.CV↗

arXiv2023

Sculpting liquid crystal skyrmions with external flows

We investigate, using experiments and numerical simulations, the distortions and the alignment of skyrmions in the liquid crystal under external flows for a range of average flow velocities. The simulations are based on the Landau-de Gennes $Q$ tensor theory both for isolated as well as for systems with many skyrmions. We found striking flow-driven elongation of an isolated skyrmion and flow alignment of skyrmions in the many-skyrmion system, both of which are also observed in the experiments. In the simulations, particular attention was given to the dissipation rate and to the various dissipation channels for a single skyrmion under external flow. This analysis provides insight on the observed scaling regime of the elongation of isolated flowing skyrmions and revealed a surprising plastic response at very short times, which may be relevant in applications based on the alignment of soft structures such as liquid crystal skyrmions.

cond-mat.soft↗

arXiv2022

Self-supervised Transformer for Deepfake Detection

The fast evolution and widespread of deepfake techniques in real-world scenarios require stronger generalization abilities of face forgery detectors. Some works capture the features that are unrelated to method-specific artifacts, such as clues of blending boundary, accumulated up-sampling, to strengthen the generalization ability. However, the effectiveness of these methods can be easily corrupted by post-processing operations such as compression. Inspired by transfer learning, neural networks pre-trained on other large-scale face-related tasks may provide useful features for deepfake detection. For example, lip movement has been proved to be a kind of robust and good-transferring highlevel semantic feature, which can be learned from the lipreading task. However, the existing method pre-trains the lip feature extraction model in a supervised manner, which requires plenty of human resources in data annotation and increases the difficulty of obtaining training data. In this paper, we propose a self-supervised transformer based audio-visual contrastive learning method. The proposed method learns mouth motion representations by encouraging the paired video and audio representations to be close while unpaired ones to be diverse. After pre-training with our method, the model will then be partially fine-tuned for deepfake detection task. Extensive experiments show that our self-supervised method performs comparably or even better than the supervised pre-training counterpart.

cs.CV↗

arXiv2021

Intelligent Online Selling Point Extraction for E-Commerce Recommendation

In the past decade, automatic product description generation for e-commerce have witnessed significant advancement. As the services provided by e-commerce platforms become diverse, it is necessary to dynamically adapt the patterns of descriptions generated. The selling point of products is an important type of product description for which the length should be as short as possible while still conveying key information. In addition, this kind of product description should be eye-catching to the readers. Currently, product selling points are normally written by human experts. Thus, the creation and maintenance of these contents incur high costs. These costs can be significantly reduced if product selling points can be automatically generated by machines. In this paper, we report our experience developing and deploying the Intelligent Online Selling Point Extraction (IOSPE) system to serve the recommendation system in the JD.com e-commerce platform. Since July 2020, IOSPE has become a core service for 62 key categories of products (covering more than 4 million products). So far, it has generated more than 0.1 billion selling points, thereby significantly scaling up the selling point creation operation and saving human labour. These IOSPE generated selling points have increased the click-through rate (CTR) by 1.89\% and the average duration the customers spent on the products by more than 2.03\% compared to the previous practice, which are significant improvements for such a large-scale e-commerce platform.

cs.IR↗

arXiv2021

Nonreversible Markov chain Monte Carlo algorithm for efficient generation of Self-Avoiding Walks

We introduce an efficient nonreversible Markov chain Monte Carlo algorithm to generate self-avoiding walks with a variable endpoint. In two dimensions, the new algorithm slightly outperforms the two-move nonreversible Berretti-Sokal algorithm introduced by H.~Hu, X.~Chen, and Y.~Deng in \cite{old}, while for three-dimensional walks, it is 3--5 times faster. The new algorithm introduces nonreversible Markov chains that obey global balance and allows for three types of elementary moves on the existing self-avoiding walk: shorten, extend or alter conformation without changing the walk's length.

cond-mat.stat-mech↗

arXiv2021

Multi-attentional Deepfake Detection

Face forgery by deepfake is widely spread over the internet and has raised severe societal concerns. Recently, how to detect such forgery contents has become a hot research topic and many deepfake detection methods have been proposed. Most of them model deepfake detection as a vanilla binary classification problem, i.e, first use a backbone network to extract a global feature and then feed it into a binary classifier (real/fake). But since the difference between the real and fake images in this task is often subtle and local, we argue this vanilla solution is not optimal. In this paper, we instead formulate deepfake detection as a fine-grained classification problem and propose a new multi-attentional deepfake detection network. Specifically, it consists of three key components: 1) multiple spatial attention heads to make the network attend to different local parts; 2) textural feature enhancement block to zoom in the subtle artifacts in shallow features; 3) aggregate the low-level textural feature and high-level semantic features guided by the attention maps. Moreover, to address the learning difficulty of this network, we further introduce a new regional independence loss and an attention guided data augmentation strategy. Through extensive experiments on different datasets, we demonstrate the superiority of our method over the vanilla binary classifier counterparts, and achieve state-of-the-art performance.

cs.CV↗

arXiv2021

Improved Image Matting via Real-time User Clicks and Uncertainty Estimation

Image matting is a fundamental and challenging problem in computer vision and graphics. Most existing matting methods leverage a user-supplied trimap as an auxiliary input to produce good alpha matte. However, obtaining high-quality trimap itself is arduous, thus restricting the application of these methods. Recently, some trimap-free methods have emerged, however, the matting quality is still far behind the trimap-based methods. The main reason is that, without the trimap guidance in some cases, the target network is ambiguous about which is the foreground target. In fact, choosing the foreground is a subjective procedure and depends on the user's intention. To this end, this paper proposes an improved deep image matting framework which is trimap-free and only needs several user click interactions to eliminate the ambiguity. Moreover, we introduce a new uncertainty estimation module that can predict which parts need polishing and a following local refinement module. Based on the computation budget, users can choose how many local parts to improve with the uncertainty guidance. Quantitative and qualitative results show that our method performs better than existing trimap-free methods and comparably to state-of-the-art trimap-based methods with minimal user effort.

cs.CV↗

arXiv2020

Modified Stokes-Einstein Relation for Small Brownian Particles

The Stokes-Einstein (SE) relation has been widely applied to quantitatively describe the Brownian motion. Notwithstanding, here we show that even for a simple fluid, the SE relation may not be completely applicable. Namely, although the SE relation could be a good approximation for a large enough Brownian particle, we find that it induces significant error for a smaller Brownian particle, and the error increases with the decrease of the Brownian particle's size, till the SE relation fails completely when the size of Brownian particle is comparable with that of a fluid molecule. The cause is rooted in the fact that the kinetic and the hydrodynamic effects depend on the size of the Brownian particle differently. By excluding the kinetic contribution to the diffusion coefficient, we propose a revised Stokes-Einstein relation and show that it expands significantly the applicable range.

cond-mat.stat-mech↗

arXiv2020

Distribution-Preserving Steganography Based on Text-to-Speech Generative Models

Steganography is the art and science of hiding secret messages in public communication so that the presence of the secret messages cannot be detected. There are two distribution-preserving steganographic frameworks, one is sampling-based and the other is compression-based. The former requires a perfect sampler which yields data following the same distribution, and the latter needs explicit distribution of generative objects. However, these two conditions are too strict even unrealistic in the traditional data environment, e.g. the distribution of natural images is hard to seize. Fortunately, generative models bring new vitality to distribution-preserving steganography, which can serve as the perfect sampler or provide the explicit distribution of generative media. Take text-to-speech generation task as an example, we propose distribution-preserving steganography based on WaveGlow and WaveNet, which corresponds to the former two categories. Steganalysis experiments and theoretical analysis are conducted to demonstrate that the proposed methods can preserve the distribution.

cs.MM↗

arXiv2019

Anisotropic thermalization propelled motor

Brownian motors and Feymann ratchets have been intensively studied in the past decades due to their significance to the foundation of statistical physics. In this work we propose a new type of Brownian motor, i.e., a self-driven motor that only utilizes the temperature difference inside and outside the motor. The motor is a container with asymmetric geometry; when filling of gas, directional motion occurs if its temperature is different from the environment. The essentially new mechanism is that the asymmetric geometry of the container may induce anisotropic thermalization, which results in a density gradient and propels the motor. The directional motion ceases untill the density gradient disappears as the inside temperature approaches the environment temperature. The same mechanism is also applied to design self-driven rotators. Possible experimental realizations are discussed.

cond-mat.stat-mech↗

arXiv2018

An $O(N)$ Sorting Algorithm: Machine Learning Sort

We propose an $O(N\cdot M)$ sorting algorithm by Machine Learning method, which shows a huge potential sorting big data. This sorting algorithm can be applied to parallel sorting and is suitable for GPU or TPU acceleration. Furthermore, we discuss the application of this algorithm to sparse hash table.

cs.LG↗

arXiv2018

Description of Brownian motion including both kinetic and hydrodynamic effects

It is known that a full description of Brownian motion in the entire course of time should incorporate both kinetic and hydrodynamic effects, but a formula accounts for both effects has been established only in three dimension and only for the limiting case when Brownian particles are much bigger and heavier than fluid particles. Nevertheless, for applications, it is important to consider small Brownian particles (sometimes even smaller than fluid particles). In these cases, unfortunately, only formulae in the short (long) time limit considering exclusively kinetic (hydrodynamic) effect are available. Here we derive a general solution to both two- and three-dimensional Brownian motion. Our solution is applicable to a wide spectrum of suspended particles from being smaller than the fluid particles to usual Brownian particles. Our analytical results are well corroborated by the numerical simulations.

cond-mat.stat-mech↗

arXiv2017

The Fourier heat conduction as a strong kinetic effect

For an one-dimensional (1D) momentum conserving system, intensive studies have shown that generally its heat current autocorrelation function (HCAF) tends to decay in a power-law manner and results in the breakdown of the Fourier heat conduction law in the thermodynamic limit. This has been recognized to be a dominant hydrodynamic effect. Here we show that, instead, the kinetic effect can be dominant in some cases and leads to the Fourier law. Usually the HCAF undergoes a fast decaying kinetic stage followed by a long, slowly decaying hydrodynamic tail. In a finite range of the system size, we find that whether the system follows the Fourier law depends on whether the kinetic stage dominates. Our study is illustrated by the 1D diatomic gas model, with which the HCAF is derived analytically and verified numerically by molecular dynamics simulations.

cond-mat.stat-mech↗