arXiv subjects

Qiang Deng

Publications and source records attributed to Qiang Deng.

18 recordsLinked to original sources

arXiv2026

LoViF 2026 The First Challenge on Weather Removal in Videos

This paper presents a review of the LoViF 2026 Challenge on Weather Removal in Videos. The challenge encourages the development of methods for restoring clean videos from inputs degraded by adverse weather conditions such as rain and snow, with an emphasis on achieving visually plausible and temporally consistent results while preserving scene structure and motion dynamics. To support this task, we introduce a new short-form WRV dataset tailored for video weather removal. It consists of 18 videos 1,216 synthesized frames paired with 1,216 real-world ground-truth frames at a resolution of 832 x 480, and is split into training, validation, and test sets with a ratio of 1:1:1. The goal of this challenge is to advance robust and realistic video restoration under real-world weather conditions, with evaluation protocols that jointly consider fidelity and perceptual quality. The challenge attracted 37 participants and received 5 valid final submissions with corresponding fact sheets, contributing to progress in weather removal for videos. The project is publicly available at https://www.codabench.org/competitions/13462/.

cs.CV↗

arXiv2022

Attribute Controllable Beautiful Caucasian Face Generation by Aesthetics Driven Reinforcement Learning

In recent years, image generation has made great strides in improving the quality of images, producing high-fidelity ones. Also, quite recently, there are architecture designs, which enable GAN to unsupervisedly learn the semantic attributes represented in different layers. However, there is still a lack of research on generating face images more consistent with human aesthetics. Based on EigenGAN [He et al., ICCV 2021], we build the techniques of reinforcement learning into the generator of EigenGAN. The agent tries to figure out how to alter the semantic attributes of the generated human faces towards more preferable ones. To accomplish this, we trained an aesthetics scoring model that can conduct facial beauty prediction. We also can utilize this scoring model to analyze the correlation between face attributes and aesthetics scores. Empirically, using off-the-shelf techniques from reinforcement learning would not work well. So instead, we present a new variant incorporating the ingredients emerging in the reinforcement learning communities in recent years. Compared to the original generated images, the adjusted ones show clear distinctions concerning various attributes. Experimental results using the MindSpore, show the effectiveness of the proposed method. Altered facial images are commonly more attractive, with significantly improved aesthetic levels.

cs.CV↗

arXiv2022

Aesthetic Language Guidance Generation of Images Using Attribute Comparison

With the vigorous development of mobile photography technology, major mobile phone manufacturers are scrambling to improve the shooting ability of equipments and the photo beautification algorithm of software. However, the improvement of intelligent equipments and algorithms cannot replace human subjective photography technology. In this paper, we propose the aesthetic language guidance of image (ALG). We divide ALG into ALG-T and ALG-I according to whether the guiding rules are based on photography templates or guidance images. Whether it is ALG-T or ALG-I, we guide photography from three attributes of color, lighting and composition of the images. The differences of the three attributes between the input images and the photography templates or the guidance images are described in natural language, which is aesthetic natural language guidance (ALG). Also, because of the differences in lighting and composition between landscape images and portrait images, we divide the input images into landscape images and portrait images. Both ALG-T and ALG-I conduct aesthetic language guidance respectively for the two types of input images (landscape images and portrait images).

cs.CV↗

arXiv2022

Aesthetic Attribute Assessment of Images Numerically on Mixed Multi-attribute Datasets

With the continuous development of social software and multimedia technology, images have become a kind of important carrier for spreading information and socializing. How to evaluate an image comprehensively has become the focus of recent researches. The traditional image aesthetic assessment methods often adopt single numerical overall assessment scores, which has certain subjectivity and can no longer meet the higher aesthetic requirements. In this paper, we construct an new image attribute dataset called aesthetic mixed dataset with attributes(AMD-A) and design external attribute features for fusion. Besides, we propose a efficient method for image aesthetic attribute assessment on mixed multi-attribute dataset and construct a multitasking network architecture by using the EfficientNet-B0 as the backbone network. Our model can achieve aesthetic classification, overall scoring and attribute scoring. In each sub-network, we improve the feature extraction through ECA channel attention module. As for the final overall scoring, we adopt the idea of the teacher-student network and use the classification sub-network to guide the aesthetic overall fine-grain regression. Experimental results, using the MindSpore, show that our proposed method can effectively improve the performance of the aesthetic overall and attribute assessment.

cs.CV↗

arXiv2020

Cognitive Representation Learning of Self-Media Online Article Quality

The automatic quality assessment of self-media online articles is an urgent and new issue, which is of great value to the online recommendation and search. Different from traditional and well-formed articles, self-media online articles are mainly created by users, which have the appearance characteristics of different text levels and multi-modal hybrid editing, along with the potential characteristics of diverse content, different styles, large semantic spans and good interactive experience requirements. To solve these challenges, we establish a joint model CoQAN in combination with the layout organization, writing characteristics and text semantics, designing different representation learning subnetworks, especially for the feature learning process and interactive reading habits on mobile terminals. It is more consistent with the cognitive style of expressing an expert's evaluation of articles. We have also constructed a large scale real-world assessment dataset. Extensive experimental results show that the proposed framework significantly outperforms state-of-the-art methods, and effectively learns and integrates different factors of the online article quality assessment.

cs.CL↗

arXiv2020

Weak Supervision for Fake News Detection via Reinforcement Learning

Today social media has become the primary source for news. Via social media platforms, fake news travel at unprecedented speeds, reach global audiences and put users and communities at great risk. Therefore, it is extremely important to detect fake news as early as possible. Recently, deep learning based approaches have shown improved performance in fake news detection. However, the training of such models requires a large amount of labeled data, but manual annotation is time-consuming and expensive. Moreover, due to the dynamic nature of news, annotated samples may become outdated quickly and cannot represent the news articles on newly emerged events. Therefore, how to obtain fresh and high-quality labeled samples is the major challenge in employing deep learning models for fake news detection. In order to tackle this challenge, we propose a reinforced weakly-supervised fake news detection framework, i.e., WeFEND, which can leverage users' reports as weak supervision to enlarge the amount of training data for fake news detection. The proposed framework consists of three main components: the annotator, the reinforced selector and the fake news detector. The annotator can automatically assign weak labels for unlabeled news based on users' reports. The reinforced selector using reinforcement learning techniques chooses high-quality samples from the weakly labeled data and filters out those low-quality ones that may degrade the detector's prediction performance. The fake news detector aims to identify fake news based on the news content. We tested the proposed framework on a large collection of news articles published via WeChat official accounts and associated user reports. Extensive experiments on this dataset show that the proposed WeFEND model achieves the best performance compared with the state-of-the-art methods.

cs.SI↗

arXiv2019

Turbulent lithosphere deformation in the Tibetan Plateau

In this work, we show that the Tibetan Plateau deformation demonstrates a turbulence-like statistics, e.g., spatial invariance cross continuous scales. A dual-power-law behavior is evident to show the existence of two possible conversation laws for the enstrophy-like cascade on the range $500\lesssim r\lesssim 2,000\,\si{km}$ and kinetic-energy-like cascade on the range $50\lesssim r\lesssim 500\,\si{km}$. The measured second-order structure-function scaling exponents $ζ(2)$ are similar with the counterpart of the Fourier scaling exponents observed in the atmosphere, where in the latter case the earth rotation is relevant. The turbulent statistics observed here for nearly zero Reynolds number flow is favor to be interpreted by the geostrophic turbulence theory. Moreover, the intermittency correction is recognized with an intensity to be close to the one of the hydrodynamic turbulence of high Reynolds number turbulent flows, implying a universal scaling feature of very different turbulent flows. Our results not only shed new light on the debate regarding the mechanism of the Tibetan Plateau deformation, but also lead to new challenge for the geodynamic modelling using Newton or non-Newtonian model that the observed turbulence-like features have to be taken into account.

physics.flu-dyn↗

arXiv2017

Collective vortex pinning and crossover between second order to first order transition in optimally doped Ba1-xKxBiO3 single crystals

Measurements on magnetization and relaxation have been carried out on an optimally doped Ba$_{0.59}$K$_{0.47}$BiO$_{3+δ}$ single crystal with $T_c$ = 31.3 K. Detailed analysis is undertaken on the data. Both the dynamical relaxation and conventional relaxation have been measured leading to the self-consistent determination of the magnetization relaxation rate. It is found that the data are well described by the collective pinning model leading to the glassy exponent of about $μ\approx$ 1.64 to 1.68 with the magnetic fields of 1 and 3 T. The analysis based on Maley's method combining with the conventional relaxation data allows us to determine the current dependent activation energy $U$ which yields a $μ$ value of about 1.23 to 1.29 for the magnetic fields of 1 and 3 T. The second magnetization peaks appear in wide temperature region from 2 K to 24 K. The separation between the second peak field and the irreversibility field becomes narrow when temperature is increased. When the two fields are close to each other, we find that the second peak evolves into a step like transition of magnetization, suggesting a crossover from the second order to first order transition. Finally, we present a vortex phase diagram and demonstrate that the vortex dynamics in Ba$_{1-x}$K$_x$BiO$_3$ can be used as a model system for studying the collective vortex pining.

cond-mat.supr-con↗

arXiv2016

Multi-band superconductivity and large anisotropy in FeS crystals

By using a hydrothermal method, we have successfully grown crystals of the newly discovered superconductor FeS, which has an isostructure of the iron based superconductor FeSe. The superconductivity appears at about 4.5K, as revealed by both resistive and magnetization measurements. It is found that the upper critical field is relatively low, with however an rather large anisotropy $Γ=[(dH_{c2}^{ab}/dT)/(dH_{c2}^{c}/dT)]_{T_c}\approx5.8$. A huge magnetoresistivity (290$\%$ at 9T and 10K, ${H}$ $\parallel$ c-axis) together with a non-linear behavior of Hall resistivity vs. external field are observed. A two-band model is applied to fit the magnetoresistance and non-linear transverse resistivity, yielding the basic parameters of the electron and hole bands.

cond-mat.supr-con↗

arXiv2015

Magnetization relaxation, critical current density and vortex dynamics in a Ba$_{0.66}$K$_{0.32}$BiO$_{3+δ}$ single crystal

We have conducted extensive investigations on the magnetization and its dynamical relaxation on a Ba$_{0.66}$K$_{0.32}$BiO$_{3+δ}$ single crystal. It is found that the magnetization relaxation rate is rather weak compared with that in the cuprate superconductors, indicating a higher collective vortex pinning potential (or activation energy), although the intrinsic pinning potential $U_\mathrm{c}$ is weaker. Detailed analysis leads to the following discoveries: (1) A second-peak effect on the magnetization-hysteresis-loop was observed in a very wide temperature region, ranging from 2K to 24K. Its general behavior looks like that in YBa$_2$Cu$_3$O$_7$; (2) Associated with the second peak effect, the magnetization relaxation rate is inversely related to the transient superconducting current density $J_\mathrm{s}$ revealing a quite general and similar mechanism for the second peak effect in many high temperature superconductors; (3) A detailed analysis based on the collective creep model reveals a large glassy exponent $μ$ and a small intrinsic pinning potential $U_\mathrm{c}$; (4) Investigation on the volume pinning force density shows that the data can be scaled to the formula $F_{p}\propto b^p(1-b)^q$ with $p=2.79$ and $q=3.14$, here $b$ is the reduced magnetic field to the irreversible magnetic field. The maximum normalized pinning force density appears near $b\approx0.47$. Finally, a vortex phase diagram is drawn for showing the phase transitions or crossovers between different vortex phases.

cond-mat.supr-con↗

arXiv2015

Anisotropic Electronic Mobilities in the Nematic State of the Parent Phase NaFeAs

Hall effect and magnetoresistance have been measured on single crystals of the parent phase NaFeAs under a uniaxial pressure. Although significant difference of the in-plane resistivity $ρ_{xx}(I\parallel a)$ and $ρ_{xx}(I\parallel b)$ with the uniaxial pressure along $b$-axis was observed, the transverse resistivity $ρ_{xy}$ shows a surprisingly isotropic behavior. Detailed analysis reveals that the Hall coefficient $R_\mathrm{H}$ measured in the two orthogonal configurations ($I\parallel a$-axis and $I\parallel b$-axis) coincide very well and exhibit a deviation from the high temperature background at around the structural transition temperature $T_{\mathrm{s}}$. Furthermore, the magnitude of $R_\mathrm{H}$ increases remarkably below the structural transition temperature. This enhanced Hall coefficient is accompanied by the non-linear transverse resistivity versus magnetic field and enhanced magnetoresistance, which can be explained very well by the two band model with anisotropic mobilities of each band. Our results together with the two band model analysis clearly show that the anisotropic in-plane resistivity in the nematic state is closely related to the distinct quasiparticle mobilities when they are moving parallel or perpendicular to the direction of the uniaxial pressure.

cond-mat.supr-con↗

arXiv2014

Simultaneous Vanishing of the Nematic Electronic State and the Structural Orthorhombicity in NaFe$_{1-x}$Co$_x$As Single Crystals

We have carried out in-plane resistivity measurements under a uniaxial pressure in NaFe$_{1-x}$Co$_x$As single crystals. A clear distinction of the in-plane resistivity $ρ_a$ and $ρ_b$ with the uniaxial pressure along $b$-axis was discovered in the parent and underdoped regime with the doping level up to about x=0.025$\pm$0.002. From the deviating point of $ρ_a$ and $ρ_b$, and the unique kinky structure of resistivity together with the published data we determined the temperatures for the nematic, structural and antiferromagnetic transitions. It is clearly shown that the nematic electronic state vanishes simultaneously with the structural transition. The antiferromagnetic state disappears however at a lower doping level. Our results, in combination with the data in BaFe$_{2-x}$Co$_x$As$_2$, indicate a close relationship between nematicity and superconductivity.

cond-mat.supr-con↗

arXiv2014

NSQUID arrays as conveyers of quantum information

We have considered the quantum dynamics of an array of nSQUIDs -- two-junction SQUIDs with negative mutual inductance between their two arms. Effective dual-rail structure of the array creates additional internal degree of freedom for the fluxons in the array, which can be used to encode and transport quantum information. Physically, this degree of freedom is represented by electromagnetic excitations localized on the fluxon. We have calculated the spatial profile and frequency spectrum of these excitations. Their dynamics can be reduced to two quantum states, so that each fluxons moving through the array carries with it a qubit of information. Coherence properties of such a propagating qubits in the nSQUID array are characterized by the dynamic suppression of the low-frequency decoherence due to the motion-induced spreading of the noise spectral density to a larger frequency interval.

cond-mat.mes-hall↗

arXiv2014

The spectroscopic signature of Kondo screening on single adatoms in Na(Fe0.96Co0.03Mn0.01)As

The electronic states of surface adatoms in Na(Fe0.96Co0.03Mn0.01)As have been studied by low temperature scanning tunneling spectroscopy. The spectra recorded on the adatoms display both superconducting coherence peaks and an asymmetric resonance in a larger energy scale. The Fano-type line shape of the spectra points towards a possible Kondo effect at play. The apparent energy position of the resonance peak shifts about 5 meV to the Fermi level when measured across the critical temperature, supporting that the Bogoliubov quasiparticle is responsible for the Kondo screening in the superconducting state. The tunneling spectra do not show the subgap bound states, which is explained as the weak pair breaking effect given by the weak and broad scattering potential after the Kondo screening.

cond-mat.supr-con↗

arXiv2013

In-Gap Quasiparticle Excitations Induced by Non-Magnetic Cu Impurities in Na(Fe0.96Co0.03Cu0.01)As Revealed by Scanning Tunneling Spectroscopy

The pairing mechanism in the iron pnictides remains unresolved yet. One of the central issues is the structure of the superconducting order parameter which classifies the community into two different and highly disputed camps. On one hand the picture of pairing based on the magnetic origin predicts a sign reversal gap on the electron and hole Fermi pockets, leading to the S+- pairing. On the other hand, a more conventional S++ pairing gap was suggested based on the phonon or orbital fluctuation mediated pairing. In the superconducting state, the impurities may generate a unique pattern of local density of states in space and energy, which are regarded as the fingerprints for checking the structure of the pairing gap. In this study, we successfully identified the non-magnetic and magnetic impurities in Na(Fe0.97-xCo0.03Tx)As (T=Cu, Mn) and investigated the spatial resolved scanning tunneling spectroscopy. We present clear evidence of the in-gap quasiparticle states induced by the nonmagnetic Cu impurities, giving decisive evidence of the S+- pairing. This is corroborated by the consistency between the experimental data and the first-principles calculations based on the S+- pairing gap with a scalar scattering potential.

cond-mat.supr-con↗

arXiv2013

Impurity effect and suppression to superconductivity in Na(Fe$_{0.97-x}$Co$_{0.03}$T$_x$)As (T=Cu, Mn)

We report the successful growth and the impurity scattering effect of single crystals of Na(Fe$_{0.97-x}$Co$_{0.03}$T$_x$)As (T=Cu, Mn). The temperature dependence of DC magnetization at high magnetic fields is measured for different concentrations of Cu and Mn. Detailed analysis based on the Curie-Weiss law indicates that the Cu doping weakens the average magnetic moments, while doping Mn enhances the local magnetic moments greatly, suggesting that the former may be non- or very weak magnetic impurities, and the latter give rise to magnetic impurities. However, it is found that both doping Cu and Mn will enhance the residual resistivity and suppress the superconductivity at the same rate in the low doping region, being consistent with the prediction of the S$^{\pm}$ model. For the Cu-doped system, the superconductivity is suppressed completely at a residual resistivity $ρ_0$ = 0.87 m$Ω$ cm at which a strong localization effect is observed. However, in the case of Mn doping, the behavior of suppression to \emph{T}$_{c}$ changes from a fast speed to a slow one and keeps superconductive even up to a residual resistivity of 2.86 m$Ω$ cm. Clearly the magnetic Mn impurities are even not as detrimental as the non- or very weak magnetic Cu impurities to superconductivity in the high doping regime.

cond-mat.supr-con↗

arXiv2013

Decoherence induced deformation of the ground state in adiabatic quantum computation

Despite more than a decade of research on adiabatic quantum computation (AQC), its decoherence properties are still poorly understood. Many theoretical works have suggested that AQC is more robust against decoherence, but a quantitative relation between its performance and the qubits' coherence properties, such as decoherence time, is still lacking. While the thermal excitations are known to be important sources of errors, they are predominantly dependent on temperature but rather insensitive to the qubits' coherence. Less understood is the role of virtual excitations, which can also reduce the ground state probability even at zero temperature. Here, we introduce normalized ground state fidelity as a measure of the decoherence-induced deformation of the ground state due to virtual transitions. We calculate the normalized fidelity perturbatively at finite temperatures and discuss its relation to the qubits' relaxation and dephasing times, as well as its projected scaling properties.

cond-mat.mes-hall↗

arXiv2013

Influence of microstructure on superconductivity in K$_x$Fe$_{2-y}$Se$_2$ and evidence for a new parent phase K$_2$Fe$_7$Se$_8$

The search for new superconducting materials has been spurred on by the discovery of iron-based superconductors whose structure and composition is qualitatively different from the cuprates. The study of one such material, K$_x$Fe$_{2-y}$Se$_2$ with a critical temperature of 32 K, is made more difficult by the fact that it separates into two phases: a dominant antiferromagnetic insulating phase with a K$_2$Fe$_4$Se$_5$ structure, and a minority superconducting phase whose precise structure is as yet unclear. Here we perform electrical and magnetization measurements, scanning electron microscopy and microanalysis, X-ray diffraction, and scanning tunnelling microscopy on K$_x$Fe$_{2-y}$Se$_2$ crystals prepared under different quenching processes to better understand the relationship between its microstructure and its superconducting phase. We identify a 3D network of superconducting filaments within this material and present evidence to suggest that the superconducting phase consists of a single Fe vacancy for every eight Fe-sites arranged in a $\sqrt{10}\times\sqrt{8}$ parallelogram structure.

cond-mat.supr-con↗